mirror of
https://github.com/block/buzz.git
synced 2026-08-18 06:50:31 +02:00
wpfleger/admin-auth-e2e
536
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 Jordan MecomandGitHub
|
885bed35ee |
fix(workflow): bind trigger author to the signed event (#4607)
This change derives `trigger_author` exclusively from the signed event pubkey. Actor tags remain available as event data but cannot override the identity used by author-sensitive workflow conditions. This removes the impersonation path without changing workflow definitions or requiring stored-data migration. ## Testing - `bin/cargo test -p buzz-workflow` at `78819df`: 154 passed, 2 Postgres-dependent tests ignored - `git diff --check origin/main...codex/security-workflow-trigger-author` Originating Buzz thread: `buzz://message?channel=3928fe05-df61-4b5d-b9c7-d623b9b10ea1&id=3c6c02312f763fbe0d2bfc33a6c1a362f91d0354f3d18b039cf7a0558c1439d1` Signed-off-by: Jordan Mecom <jm@squareup.com> |
||
|
|
997b8caaa4 |
fix(git): revoke access for banned relay members (#4608)
This change rechecks the durable community ban in the shared Git HTTP authentication path for advertise, fetch, and push requests. A banned member is denied even if repository-channel membership still exists, and restriction lookup errors fail closed. The additional database lookup happens on every Git HTTP request so access revocation does not depend on stale session state. The check also cascades to the NIP-OA owner. Git accepts NIP-OA attestations on the NIP-98 token, so an agent key can act for its owner — without the cascade, a banned human would keep clone and push access through any agent key. This mirrors the NIP-42 gate in `handlers::auth`: either principal's ban denies the request. The check runs inside the `GitAuth` extractor, so all three Git routes inherit it. ## Testing - `git diff --check origin/main...codex/security-ban-revokes-git` - Rebased onto `origin/main` at `5c98932` - `cargo test -p buzz-relay --lib sec005_read_gate_tests`: 8 passed, 7 ignored (Postgres) - `cargo clippy -p buzz-relay --all-targets -- -D warnings` and `cargo fmt --check`: clean Pure tests cover the decision table (agent ban, inherited owner ban, no attestation). Postgres-gated tests cover the wiring: the real ban row, a live `compute_auth_tag` attestation, and the 503 fail-closed path. **Not yet verified:** the three Postgres-gated tests compile and skip but have not been run — no local Postgres, and CI does not run `--ignored`. They need `cargo test -p buzz-relay --lib sec005_read_gate_tests -- --ignored` against a migrated dev database. Originating Buzz thread: `buzz://message?channel=3928fe05-df61-4b5d-b9c7-d623b9b10ea1&id=3c6c02312f763fbe0d2bfc33a6c1a362f91d0354f3d18b039cf7a0558c1439d1` --------- Signed-off-by: Jordan Mecom <jm@squareup.com> Signed-off-by: Eli Foster <efoster@squareup.com> Co-authored-by: Eli Foster <efoster@squareup.com> Co-authored-by: Claude Opus 5 <noreply@anthropic.com> |
||
|
|
8a7eb8d3d7 |
fix(agent): recover from unsupported image input instead of poisoning the turn (#4896)
## Problem
`buzz-dev-mcp` advertises `view_image` to every agent regardless of
whether the session's model accepts images. When a text-only model (e.g.
DeepSeek V4 Flash) takes the bait, the image lands in session history
and every subsequent LLM request 404s with `No endpoints found that
support image input`. The error was classified as `LlmModelNotFound` and
propagated fatally out of the turn loop — history stays poisoned,
buzz-acp retries the batch with exponential backoff, and the session
burns its entire clock doing no work. In a recent trial run, **all 57
trials that called `view_image` on a text-only model died this way; none
recovered.**
## Fix
Capability-gating the advertised tool isn't reliable — there is no
image-capability metadata at the agent layer across providers. Instead,
recover at the turn loop:
- **Typed error**: new `AgentError::UnsupportedImageInput`, classified
narrowly on the exact provider phrase `No endpoints found that support
image input` on both the generic 404 path and OpenRouter's 404 path.
Unknown-model 404s and OpenRouter parameter-routing 404s keep their
existing classifications. No deterministic retry.
- **In-turn recovery**: on this error, `RunCtx::run` strips every image
block from history — keeping the tool result (and therefore
tool-call/result pairing) intact — marks the result `is_error`, appends
actionable model-facing guidance ("The current model does not support
image input. The image was removed from conversation history so this
turn can continue. Use a text-based inspection tool…"), and continues
the same turn. Base64 never replays again.
- **Loop guard**: recovery only fires when at least one image was
removed; if the provider says "image" and history has none, the error
propagates as before.
## Tests
- Unit: phrase classification (typed, not retried; unknown-model 404
unaffected), idempotent image-to-error history mutation preserving call
IDs and text.
- End-to-end (`fake_llm.rs` + `fake_mcp.rs`): tool call → MCP image
result → 404 unsupported-image → same-turn recovery. Captured requests
prove round 2 carried the image, round 3 replays no image, carries the
guidance text, preserves pairing, and ends `end_turn`.
- Loop guard: typed unsupported-image error with **no** image in history
fails after exactly one provider request instead of spinning —
mutation-testing showed deleting the `removed == 0` guard survived the
suite, and `max_rounds` defaults to unlimited in production, so this
branch needed direct coverage.
Verified at `a210305019b33d5f56677b4c82bab79e4ac52d24`: `cargo test -p
buzz-agent` (full package, 381 unit + all integration suites) green;
`clippy --all-targets -D warnings` green; `fmt --check` green; pre-push
hooks (rust-tests, desktop-tauri-checks, branch-skew) green.
**Scope of the classification guarantee**: the classifier runs in the
shared `post()` (which Anthropic and OpenAI paths route through) and in
`openrouter_post()` — i.e., every 404 path in `llm.rs`. It only runs on
404 responses; providers that reject images with a different status
(e.g. a 400) are out of scope for this PR — see the review-comment
discussion for why broadening the phrase list alone would not cover
them.
Authored by Wren, loop-guard test by Sami, reviewed by Eva.
---------
Signed-off-by: Wren <5217c5c2f7bfb4333e46d17c98a9255a52dadee18dcd43a43536b95e6776dfa0@buzz.block.builderlab.xyz>
Signed-off-by: Sami <f4a42a97e594b77bdbd8ee35191c8b28a94a4cb871d96f32921558275421fb68@buzz.block.builderlab.xyz>
Co-authored-by: Wren <5217c5c2f7bfb4333e46d17c98a9255a52dadee18dcd43a43536b95e6776dfa0@buzz.block.builderlab.xyz>
Co-authored-by: Sami <f4a42a97e594b77bdbd8ee35191c8b28a94a4cb871d96f32921558275421fb68@buzz.block.builderlab.xyz>
|
||
|
|
067c085f37 |
Define private managed agent wire protocol (#4593)
## Summary - reserve kind `30179` for owner-private managed-agent aggregates - define the fail-closed owner-self NIP-44 v2 envelope and versioned payload codec - bind runnable identity/configuration to complete signed `30175`/`30177` recovery projections - validate NIP-OA owner→agent attestations and reject self-attestation - document NIP-PMA authority, migration prerequisites, privacy, and deployment order - keep generic relay ingest closed until private storage and atomic aggregate CAS exist ## Safety boundary This is the inert protocol/codec slice only. It does not publish secrets, change agent authority, migrate local records, or enable kind `30179` ingestion. The relay regression test proves generic EVENT ingest still rejects the kind. The finalized migration plan adds later prerequisites for relay-private storage/CAS, runtime lease/fencing, Desktop cutover, and harness authentication. Those belong in staged follow-up PRs rather than expanding this inert foundation. ## Validation At commit `67f0ea4ebb8d3ccba3a3eb9374e89a7178913f74`: - `cargo test -p buzz-core` — 246 unit + 2 doc tests passed - `cargo test -p buzz-relay private_managed_agent_kind_remains_rejected_until_atomic_ingest_exists` — passed - push hooks: Rust tests and desktop checks passed (`2145` desktop tests passed, `14` ignored) - `cargo fmt --all -- --check` - `git diff --check` ## Review Princess Donut cleared security/data integrity with no remaining high/medium findings. Mongo cleared migration compatibility and wire grammar. The later runtime lease/fencing protocol was also adversarially cleared as a plan; implementation slices still require independent evidence before activation. Deterministic plaintext/signed-projection/auth-tag interoperability vectors remain a valuable follow-up, not an S0 merge gate; random NIP-44 ciphertext is intentionally not snapshotted. --------- Signed-off-by: Wes <wesbillman@users.noreply.github.com> Co-authored-by: Carl <c7ebe626f000404285d3686e1dc74cc07cc60a9754a150041ba132e14bd3e2ec@buzz.block.builderlab.xyz> |
||
|
|
4a2305170e |
fix: reauthenticate databricks model discovery (#4008)
## Summary - preserve Databricks catalog 401 responses as authentication failures and retry discovery exactly once after silently refreshing the rejected bearer - preserve runtime OAuth recovery: when discovery has no usable OAuth credential, `session/new` succeeds with only the trimmed configured model so the first `session/prompt` can run the existing browser PKCE flow - reject a rejected configured `DATABRICKS_TOKEN` with actionable, non-interactive guidance; static credentials cannot recover through PKCE - use the configured-model fallback for non-auth discovery failures without caching failed or fallback catalogs, so later sessions retry discovery - keep known Databricks v2 models only for authenticated empty-catalog responses and mark their provenance - resolve discovery before MCP spawn or session registration, preventing failed discovery from leaking resources or consuming session capacity - permit serialized interactive PKCE only from the explicit saved-agent model picker; passive draft discovery never opens a browser ## Runtime flow 1. OAuth discovery attempts cached credentials and silent refresh without opening a browser. 2. If no usable OAuth bearer exists, `session/new` advertises only the configured model and succeeds. 3. The first `session/prompt` uses `TokenSource::bearer()`, which may launch browser PKCE. 4. A later session retries discovery and caches only the authenticated catalog. ## Regression coverage - rejected-but-locally-fresh OAuth bearer performs one refresh and one catalog retry - OAuth mode with no cached token allows `session/new` and returns exactly the trimmed configured model - the OAuth fallback is not cached; a later authenticated session retries discovery and caches the returned catalog - rejected static tokens still reject `session/new` - failed discovery does not consume the sole session slot or spawn the supplied MCP process - Desktop interactive/passive auth intent, static-token redaction, and authenticated empty-catalog provenance ## Verification - `cargo test -p buzz-agent` - `cargo test --manifest-path desktop/src-tauri/Cargo.toml --lib commands::agent_models` - `cargo clippy --manifest-path desktop/src-tauri/Cargo.toml --all-targets -- -D warnings` - `cargo fmt --all -- --check` - `git diff --check` - full pre-push hooks ## Review Adversarial review found and drove fixes for session/MCP resource leakage, duplicate concurrent PKCE flows, sensitive error propagation, incorrect 403 reauthentication, missing discovery-level coverage, passive browser launch, and the Desktop file-size ratchet. The final follow-up preserves the existing prompt-time OAuth flow while retaining static-token rejection and pre-allocation discovery ordering. --------- Signed-off-by: npub1dccv64krpcpse5cmkzfeh998cftungyatw3djt8jwdw6g43f7fyqzzmrf7 <6e30cd56c30e030cd31bb0939b94a7c257c9a09d5ba2d92cf2735da45629f248@buzz.block.builderlab.xyz> Co-authored-by: npub1dccv64krpcpse5cmkzfeh998cftungyatw3djt8jwdw6g43f7fyqzzmrf7 <6e30cd56c30e030cd31bb0939b94a7c257c9a09d5ba2d92cf2735da45629f248@buzz.block.builderlab.xyz> |
||
|
|
a1d78f2959 |
feat: Buzz entity links — rich preview cards + in-app navigation for repos, PRs, and issues (#4695)
## Summary
Gives Buzz-hosted git entities the same "GitHub-style" chat experience
GitHub links already get: rich preview cards, real titles, and
click-through — except clicks navigate **in-app** to the Projects view
instead of a browser.
- **Spec**: `docs/buzz-entity-links.md` — link scheme, slices, and
deferred work (`buzz://project`, OS deep links, web routes).
- **Canonical `buzz://` deep links**: new
`desktop/src/shared/lib/entityLink.ts` with builders + strict parser for
`buzz://pr?id=…&owner=…&d=…`, `buzz://issue?…`, and
`buzz://repo?owner=…&d=…`, mirrored by a Rust module
(`crates/buzz-cli/src/links.rs`) with a shared golden-format test so the
two implementations can't drift.
- **Preview cards**: `linkPreview.ts` recognizes `buzz://` entity links
*and* HTTPS relay clone URLs (`{origin}/git/<pubkey>/<repo>`, the shape
agents paste today). Both normalize onto the canonical `buzz://` href,
so the two spellings of a repo dedupe to one `Buzz`-provider card
(`BuzzMark` logo) rendered by `link-preview-attachment.tsx`.
- **Title enrichment**: PR/issue cards fetch the real subject from the
relay event (`subject` tag or first content line) via
`useResolvedLinkPreviews.ts`; the cache is community-scoped and reset in
`resetCommunityState()`.
- **In-app navigation**: clicking a card or inline anchor (including
HTTPS relay clone URLs whose origin matches the active relay) routes to
the canonical `30617:<owner>:<d>` coordinate via `goProject()`
(`markdown/entityLinks.tsx`). **Merge dependency: #4671 must merge
first** — route resolution for `30617:` coordinates is implemented on
that branch (`feat/multi-repository-projects`). Entity-link and
external-anchor logic were extracted out of `markdown.tsx` to stay under
the file-size ratchet.
- **Agent side**: `buzz pr open`, `buzz issues create`, and `buzz repos
create` now return a ready-made `link` field (omitted when the relay
returns `accepted: false`), and `base_prompt.md` instructs agents to
paste it verbatim when announcing work.
## Test plan
- [x] Desktop unit tests: pass, including new `entityLink.test.mjs` and
`linkPreview.test.mjs` coverage (golden formats, malformed-link
rejection, clone-URL/`buzz://` dedupe, origin-gated anchor behavior,
label-must-win invariant, cache epoch)
- [x] Rust: `cargo test -p buzz-cli` golden-format test +
accepted/rejected link guard assertions, clippy + fmt clean
- [x] Biome + `tsc --noEmit` clean; pre-push hooks
(desktop-tauri-checks, rust-tests, desktop-test) pass
- [ ] Manual: paste a relay clone URL and a `buzz://pr` link in a
channel — verify one card each, real PR title, and in-app navigation to
the Projects view
Related: [#4671](https://github.com/block/buzz/pull/4671)
---------
Signed-off-by: Thomas Petersen <thomasp@squareup.com>
Signed-off-by: Will Pfleger <pfleger.will@gmail.com>
Co-authored-by: npub1g8493u0xfsjrvflg4n08ezd7vec99mnwzlv0qgwpr9d7gvjwhuzqx59rhw <41ea58f1e64c243627e8acde7c89be667052ee6e17d8f021c1195be4324ebf04@buzz.block.builderlab.xyz>
Co-authored-by: Will Pfleger <pfleger.will@gmail.com>
|
||
|
|
e30db7028f |
feat(projects): support multiple repositories (#4671)
## Summary - adopt the finalized NIP-MP project model so one project can enumerate and switch between multiple NIP-34 repositories - add project and repository navigation, activity summaries, existing-repository attachment, and repository access-channel management - preserve privacy-safe activation provenance for agent-authored patches, pull requests, issues, and associated commits ## Test plan - [x] Run desktop typecheck and unit tests - [x] Run focused NIP-MP, repository access, and provenance tests - [x] Run Rust formatting and desktop lint checks - [x] Run the complete pre-push suite after merging current `main` - [ ] Manually verify project creation, repository attachment, switching, and access repair on staging - [ ] Manually verify public-channel and private-agent origin labels on newly created Git activity Related: [#4695](https://github.com/block/buzz/pull/4695) --------- Signed-off-by: Thomas Petersen <thomasp@squareup.com> Signed-off-by: Will Pfleger <pfleger.will@gmail.com> Co-authored-by: npub1mn7jgtj4w2pd0g0zeuhxsa6jy6p0rewxz4kujt98my82ahfmp72sxjexk7 <dcfd242e557282d7a1e2cf2e6877522682f1e5c6156dc92ca7d90eaedd3b0f95@buzz.block.builderlab.xyz> Co-authored-by: Will Pfleger <pfleger.will@gmail.com> |
||
|
Jemiah WestermanandGitHub
|
bc9e6528a7 |
perf(relay): index channel-id lookups and skip trace-only reads (#4647)
## Problem
`SELECT id, community_id FROM channels WHERE id = ANY($1) AND deleted_at
IS NULL` is the top **Load by waits (AAS)** on the Buzz Postgres writer.
Two independent causes compound, and both are fixed here.
### 1. No index can serve it
`channels` is `PRIMARY KEY (community_id, id)`, and every secondary
index leads with `community_id`:
| Index | Columns |
|---|---|
| *(primary key)* | `(community_id, id)` |
| `idx_channels_nip29_group` | `(community_id, nip29_group_id)` |
| `idx_channels_dm_hash` | `(community_id, participant_hash)` |
| `idx_channels_community_type` | `(community_id, channel_type)` |
| `idx_channels_community_visibility` | `(community_id, visibility)` |
| `idx_channels_created_by` | `(community_id, created_by)` |
| `idx_channels_ttl_expiry` | `(ttl_deadline)` *(partial)* |
The two tenant-independent lookups carry **no `community_id` predicate**
— deliberately:
- `Db::communities_of_channels` — `WHERE id = ANY($1) AND deleted_at IS
NULL`
- `Db::community_of_channel` — `WHERE id = $1 AND deleted_at IS NULL`
That independence is load-bearing, not an oversight: projecting a row's
*true* owning community regardless of the fetch query's `WHERE` clause
is what makes `Inv_NonInterference` non-vacuous. If the fetch ever
dropped its tenant scoping, this lookup would still report the real
label and the checker would catch the mismatch.
But a composite btree is only usable when its leading column is
constrained, so neither query can use the primary key, and nothing else
leads with `id`. **Both sequentially scan `channels` on every call.**
### 2. In production the result is discarded
Both call sites feed `record_read_message_rows` /
`record_read_by_id_rows`, which call `tracer.record(...)`. Production
binds `NoopTracer` (`crates/buzz-relay/src/state.rs`), whose `record`
body is empty.
The existing guard tests `trace_state`, which is `Some` for every
well-formed request — it only goes `None` on malformed pubkey bytes. So
the scan ran on the hot read path and its output was dropped. This is
the classic eager-argument bug: `log.debug("..." + expensiveCall())`
with no `isDebugEnabled()` check.
### 3. Multiplied per filter
The non-search call site sits **inside the phase-3 per-filter loop**, so
a `REQ` carrying N filters performed N sequential scans of `channels`
before responding.
## Changes
**`Tracer::enabled()`** — a capability check on the trait (the
`isDebugEnabled()` of this seam), defaulting to `true`. `NoopTracer`
overrides it to `false`, and both emitters in `req.rs` now gate on it,
skipping the trace-only DB read entirely in production.
**`migrations/0027_channels_id_lookup_index.sql`**
```sql
CREATE INDEX IF NOT EXISTS idx_channels_id_live
ON channels (id) INCLUDE (community_id)
WHERE deleted_at IS NULL;
```
- `INCLUDE (community_id)` — both queries select exactly `(id,
community_id)`, so this is covering and can be served index-only.
- Partial on `deleted_at IS NULL` — matches both predicates exactly,
excludes soft-deleted history, and lets Postgres skip the recheck.
- **Not `UNIQUE`.** `id` alone is *not* unique in this table —
`command_executor.rs` documents that `community_of_channel(channel_id)`
is ambiguous because the same channel id can appear under more than one
community. A unique index would encode a false constraint and fail to
build on any database already holding such a pair.
Worth keeping the index even though fix #1 removes the production
caller: it still runs under conformance, and `community_of_channel` has
the same problem on its own paths.
**`schema/schema.sql`** — mirrored, since a test asserts desired-state
parity.
## Conformance is unchanged
This is the part worth reviewing closely. Under a real tracer
`enabled()` returns `true` and **every emit happens exactly as before**
— the gate only skips *building* emit inputs when nothing observes them,
never an emit that would otherwise have been made. The coverage-breach
guard stays non-vacuous.
`CountingTracer` forwards `enabled()` to its inner tracer rather than
inheriting the `true` default. Both directions matter and both fail
silently:
- inheriting `true` over a `NoopTracer` would keep the overhead this PR
removes;
- hardcoding `false` over a live tracer would suppress the emits whose
absence `EmitGuard` reports as `ImplBug` — masking real breaches behind
expected ones.
Covered by a new regression test,
`counting_tracer_delegates_enabled_to_inner`, which asserts delegation
in both directions.
## Verification
- `cargo check -p buzz-conformance -p buzz-relay` — clean
- `cargo clippy --all-targets` — clean, zero warnings
- `cargo test -p buzz-conformance` — 6/6
- `cargo test -p buzz-relay --lib conformance` — 11/11
- `cargo test -p buzz-db --lib migration` — 7/7
- `just test-unit` (pre-push) — green
Migration-count assertions in `crates/buzz-db/src/migration.rs` were
bumped 26 → 27, with content assertions for 0027 following the existing
per-migration pattern (including a guard that it never becomes
`UNIQUE`).
## Open questions for reviewers
1. **Lock strategy.** Built *without* `CONCURRENTLY`, following
migration 0004's precedent, because sqlx runs each migration inside a
transaction and `CREATE INDEX CONCURRENTLY` cannot run in one. This
takes a brief `SHARE` lock on `channels` (blocks writes, not reads) —
small relative to `events`, but an operator preferring zero
write-blocking can pre-build it by hand and `IF NOT EXISTS` makes the
migration a no-op. I could not confirm whether sqlx 0.9 supports a `--
no-transaction` directive; if it does, that may be preferable.
2. **Diagnosis is static.** This comes from reading the source, not from
`EXPLAIN` against the live database. Worth confirming with `EXPLAIN
(ANALYZE, BUFFERS)` on the writer before/after — that also sizes the win
by revealing the real table size and row counts.
3. **Expected impact** scales with average filters-per-`REQ`, which I
did not measure. `pg_stat_statements` ordered by `total_exec_time` would
confirm this query drops off the top and show whether anything else is
scanning the same way.
Signed-off-by: Jemiah Westerman <jemiah@squareup.com>
|
||
|
|
56003ebf98 |
docs(acp): explain per-channel session model in base prompt (#4729)
## Overview Agents running in Buzz have no built-in awareness that each channel is an isolated conversation context. When a human mentions work "you" are doing in another channel, the current session can misread this as its own active context and try to coordinate, re-plan, or take ownership of it — causing confusion and wasted turns. ## What changed Added a `## Session Model` section to `crates/buzz-acp/src/base_prompt.md`, inserted immediately after the opening paragraph and before `## Buzz CLI`. The section explains: - Each channel is a separate session; multiple sessions of the same agent identity may be active simultaneously. - Sessions share core memory, workspace, and relay — but not conversation context or in-flight reasoning. - Cross-channel work belongs to the owning session by default; the current session may take it over only when the human explicitly requests it. No runtime code changes. Base prompt only. --------- Signed-off-by: Will Pfleger <pfleger.will@gmail.com> Co-authored-by: npub1mn7jgtj4w2pd0g0zeuhxsa6jy6p0rewxz4kujt98my82ahfmp72sxjexk7 <dcfd242e557282d7a1e2cf2e6877522682f1e5c6156dc92ca7d90eaedd3b0f95@buzz.block.builderlab.xyz> |
||
|
|
feccf4eabc |
Polish mobile inbox and media flows (#4512)
## Summary - make mobile unread state visible with bold channel names, an animated Inbox badge, and swipe-to-toggle Inbox rows - add directional transitions for top-level mobile navigation - let mobile send while media uploads, with cancellable progress UI - normalize iOS and Android video uploads, attach poster frames, and improve native video playback ## Validation - `just mobile-check` - `just mobile-test` - `cargo test -p buzz-media` - Pixel smoke test - iPhone smoke test Desktop background uploads moved to #4522 so the two platforms can be reviewed independently. --------- Signed-off-by: kenny lopez <klopez4212@gmail.com> Signed-off-by: Tom Brow <tomb@block.xyz> Co-authored-by: leader <71e9f2c44a6932b6772caaaccda1911d010463c3e2c6c40410b8329956046801@buzz.block.builderlab.xyz> Co-authored-by: Tom Brow <tomb@block.xyz> |
||
|
|
09c86c56e5 |
fix: report agent usage per provider round, not once per turn (#4545)
## The bug buzz-agent emitted its `usage_update` notification in exactly one place: after `ctx.run()` returned. Until that moment a turn's token counters lived only in the prompt task's stack frame. **A turn killed mid-flight reported nothing at all** — the provider had already billed every round it completed, and no consumer ever saw any of it. That is not a corner case for anything that ends a turn on a clock. It is the normal case for a long-horizon benchmark run that relaunches its agent between phases. ## How big Measured against a provider's own billing ledger over one run's window: | | provider ledger | what we recorded | |---|---|---| | the relaunched lead seat | $485 / 348M tok | $98.99 / 90.3M tok | | the two seats that were not relaunched | $29.90 / 856M | $25.81 / 765M — reconciles | 97% of that run's usage rows came back all zeros, against 1–4% for comparable runs that never relaunch. In one 450-phase trial exactly 7 phases recorded any usage — and each of those carries 177k–437k input tokens, a whole session's worth landing in the one phase that happened to end gracefully. Worth being precise about what was *not* wrong, since both were plausible and both were checked: - **Not pricing.** The rates were verified against the provider's endpoints API and match what we charge. - **Not a truncation bug.** The usage files were intact and internally consistent. The tokens were never captured in the first place. ## The fix The run loop now emits a session-cumulative `usage_update` after every usage-bearing provider response, so an interrupted turn has reported everything but its single in-flight request. - **Emitting more than once per turn is already part of the contract.** buzz-acp's `UsageTracker` advances its committed baseline only at publish time, and goose behaves the same way — which is why the tracker was written to tolerate it. - **The turn-start session baseline is snapshotted into `RunCtx`** so the mid-turn figure stays *session*-cumulative. A turn-local number would be discarded by a high-water-mark consumer and lose the turn entirely; there is a test for exactly that. - **Snapshot by value, not a session handle.** The loop reports once per round, and taking the sessions lock on each would serialise concurrent sessions behind one another's provider round-trips. Nothing else advances those counters while the turn holds `busy`, so it cannot go stale. - **One shared `wire::usage_update_payload`** for both call sites, so the mid-turn and end-of-turn shapes cannot drift. A drift there would present as tokens silently vanishing, which is the failure this reporting exists to prevent. ## Why not a SIGTERM handler That was the obvious shape and it does not work. At signal time the counters are not sitting anywhere a handler could reach — they are in the turn's stack frame, and the value the handler would need has not been folded into the session yet. Making usage durable *during* the turn is what actually fixes it; once it is, a handler adds nothing beyond the in-flight request, whose cost is unknown until its response lands. ## Tests - `usage_is_reported_after_each_round_not_only_at_turn_end` — two rounds; asserts the **first** notification carries round 1's counts alone, proving it went out before round 2 returned. - `mid_turn_usage_includes_earlier_turns` — a mid-turn report must be session-cumulative, not turn-local. buzz-agent 18/18 on the `fake_llm` suite, 382 unit. `cargo fmt` / `clippy` / `cargo check --workspace --all-targets` clean. ## Scope Agent-side only, against `main`. The matching harness change — settling usage on the timeout path, which was skipped on the reasoning that an incomplete turn has nothing to flush — is **#4553**, against the benchmark branch, since that harness does not exist on `main`. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Signed-off-by: Atish Patel <atish@squareup.com> Co-authored-by: Claude Code <noreply@anthropic.com> |
||
|
|
7ff5fc3189 |
feat(acp): deliver system prompt via _meta.systemPrompt for claude-agent-acp (#4395)
`claude-agent-acp` (since v0.6.0 / PR #91) accepts `_meta.systemPrompt: {append: text}` on `session/new` to append to the adapter's native preset while keeping its tool-use prompt intact — the same non-standard extension pattern as `_session/steering` was before it was standardised. ## What changes **Rust (`crates/buzz-acp/`)** - Adds `SystemPromptTransport` enum to `acp.rs`: `Field(&str)` (ACP protocol v2, unchanged) vs `ClaudeMeta(&str)` (new `_meta.systemPrompt: {append: text}`). When both `ClaudeMeta` and `session_title` are present the two `_meta` members are merged into one object so neither clobbers the other. - Gates on exact adapter identity `@agentclientprotocol/claude-agent-acp` in `pool.rs`: `session_new_system_prompt()` routes that name to `ClaudeMeta` regardless of reported `protocolVersion` (CC declares v1). `has_system_prompt_support()` gains the same name check so user-message `[Base]`/`[System]` framing is suppressed for CC sessions. - All other paths — goose post-hoc method, protocol-v2 `Field`, legacy user-message framing — are byte-identical to before. **Desktop (`desktop/src/features/agents/ui/`)** - `agentSessionTranscript.ts`: the `session/new` extractor now checks `params._meta.systemPrompt.append` as a fallback when bare `params.systemPrompt` is absent. Bare field takes precedence. Net line count stays at 1173 (ratchet limit). - `agentSessionTranscript.test.mjs`: two new tests — one verifying the `_meta` transport produces the identical standalone card (same five sections, same `turnId: null`, same placement before the first turn) as the bare-field transport; one proving bare field wins when both transports are present. ## Gate claim `@agentclientprotocol/claude-agent-acp` implies `_meta.systemPrompt` support because the feature landed in v0.6.0 (Oct 2025, commit `ea796f3`) before the `@zed-industries/claude-code-acp` → `@agentclientprotocol/claude-agent-acp` package rename (Mar 2026, commit `b409782`). The new name is therefore a reliable capability gate; the old name falls through to the protocol-version gate (status quo, no regression). ## Tests - Rust: Claude append serialization; `_meta` coexistence with `sessionTitle`; protocol-v2 bare field byte-identical; codex/old-zed omission; claude-name support/suppression gate; old `@zed-industries` name falls through to protocol-version gate. - Desktop: `_meta` transport → identical standalone card; bare field wins over `_meta` when both present. ## Pre-existing failures `just mobile-check` and `just mobile-test` fail identically on clean `origin/main` (5 `compose_bar` / `channels_page` tests + 3 Flutter lint warnings) — not caused by this change. All other `just ci` jobs are green. Signed-off-by: Will Pfleger <pfleger.will@gmail.com> Co-authored-by: npub1mn7jgtj4w2pd0g0zeuhxsa6jy6p0rewxz4kujt98my82ahfmp72sxjexk7 <dcfd242e557282d7a1e2cf2e6877522682f1e5c6156dc92ca7d90eaedd3b0f95@buzz.block.builderlab.xyz> |
||
|
|
318fbf896e |
fix(security): bump nostr crates for RUSTSEC-2026-0225..0232 + default sprig image to published digest (#4392)
## What Two changes, both fallout/follow-up from #4289 landing: ### 1. Fix the Security job failing on main (lockfile-only) Eight RUSTSEC advisories published today against the nostr stack turned `cargo-deny check` advisories red on main ([failing run](https://github.com/block/buzz/actions/runs/30761611723/job/91533106673)). Not introduced by #4289 — the advisories landed upstream and any push to main today would have tripped them. - **RUSTSEC-2026-0225..0230** → `nostr` 0.44.6 → **0.44.7** (Debug output exposing NIP-46/NIP-60 credentials; wallet parsers accepting unauthenticated events; NIP-44/NIP-04/NIP-98 resource exhaustion; NIP-50 empty-filter panic) - **RUSTSEC-2026-0231..0232** → `nostr-relay-pool` 0.44.2 (root) / 0.44.1 (tauri) → **0.44.3** (auth-challenge memory exhaustion; processing of unverified relay events) Both workspace lockfiles bumped (`Cargo.lock`, `desktop/src-tauri/Cargo.lock`). No manifest changes. ### 2. Default the desktop GUI's sprig image to the published `ghcr.io/block/buzz-sprig` The first main-push after #4289 published the image publicly (package created 18:44Z, visibility `public`). The `config_schema()`'s `image` property now carries a `default`: ``` ghcr.io/block/buzz-sprig:sha-6530b58@sha256:17facfc7608d8ddb33bc056c9aaba1098f4ef6abe5655702fbfd7584d1f74d76 ``` **Why tag+digest, not tag:** the backend deliberately rejects tag-only references — the pod runs with the agent's nsec and tags are mutable pointers (`image.rs` §Image). The tag+digest form keeps the human-traceable `sha-6530b58` while the digest does the pinning; `image::parse` already normalizes it to the tagless canonical form, so create-intent fingerprints are identical to the bare-digest spelling. The digest is the **multi-arch manifest-list digest** (amd64+arm64), resolved via `docker buildx imagetools inspect`. **This is a UI prefill, not a baked fallback:** `image` stays in the schema's `required` list, an empty value still fails closed with a named field, and the desktop submits the value explicitly in `provider_config` (the `WhereToRunSection` probe seeds `providerConfig` from schema defaults) — so deploy fingerprints never depend on compiled-in provider state, and the spec's §K8s pod-reconciliation concern about baked-default divergence is not engaged. Module prose that said "no published image exists yet" is updated to match reality. No desktop code changes needed: the form already prefills from `properties[*].default` and submits seeded defaults. ## Testing - `cargo-deny check` at head: **advisories ok, bans ok, licenses ok, sources ok** (was: advisories FAILED) - `cargo test -p buzz-backend-kubernetes`: **158 passed** (154 lib + 4 wire), including new `schema_default_image_round_trips_through_parse` pinning the constant + its normalization, and the wire `info` test now asserting the default is present in the provider's real stdout response - Live provider probe: `{"op":"info"}` against the built binary returns the default in `config_schema.properties.image.default` with `required` unchanged (`["namespace","image"]`) - Full workspace test suite via pre-push hook: green (earlier direct `cargo test --workspace` run: sole failure was `api::mesh_demo::demo_join_forwarded_arm_round_trips_echo`, the documented pre-existing main flake — unrelated, fails on base) - Image existence verified against GHCR: `docker buildx imagetools inspect ghcr.io/block/buzz-sprig:sha-6530b58` resolves to the pinned manifest-list digest with linux/amd64 + linux/arm64 manifests --------- Signed-off-by: npub1qyvc0c5kl4gqv2fd97fsk46tu378sqgy35vc83rvgfwne90sel7s0ed67d <011987e296fd5006292d2f930b574be47c7801048d1983c46c425d3c95f0cffd@buzz.block.builderlab.xyz> Co-authored-by: npub1qyvc0c5kl4gqv2fd97fsk46tu378sqgy35vc83rvgfwne90sel7s0ed67d <011987e296fd5006292d2f930b574be47c7801048d1983c46c425d3c95f0cffd@buzz.block.builderlab.xyz> |
||
|
TylerGitHubnpub1mprnacetjua2xx3p5eddmhxyk6wv929ymm5py8kd2xfxurxahspqqlgytanpub17jjz49l9jjmhhk7cac63j8yt9z555n9cw8vk7v5jz4vzw4ppld5qgj57ccnpub1qyvc0c5kl4gqv2fd97fsk46tu378sqgy35vc83rvgfwne90sel7s0ed67dDawnnpub1jh9wn95s0472h86ahapupaf7m6kx4v9sx2n0atj2hltcfer8k06s5n3pyfnpub1t2tgm7d8f995uqvmnm8h88sg3wnpp9a5xysjf6dg3tjmgt3ltulqdp8ehr |
6530b58a61 |
feat(k8s): Kubernetes backend plugin + desktop deploy path (#4289)
# Kubernetes backend plugin (crates/buzz-backend-kubernetes) + desktop deploy path Implements docs/remote-agents.md (merged @ |
||
|
Tal WeissandGitHub
|
fc598f5f8d |
fix(git): allow deleting the default branch (#4297)
Tal here, human. Trying to help. This bug bugged me... ## Summary A repository's first branch becomes its symbolic `HEAD`, and Git's bare-repository default rejects deleting that branch even when another branch survives. This change: - sets `receive.denyDeleteCurrent=ignore` only for the ephemeral `git receive-pack` process - preserves the existing server-side `core.hooksPath` override and authorization hook - lets the existing CAS publication logic select a surviving branch as the next manifest `HEAD` - adds regression coverage using a real stateless `git receive-pack` request and a manifest HEAD-selection test This lets users replace an accidental default branch without deleting the object-storage manifest pointer. ### Related issue Fixes #3572 ### Testing - `cargo test -p buzz-relay api::git::` (128 passed, 5 ignored) - `just ci` - live E2E roundtrip against a release relay with PostgreSQL, Redis, and MinIO: - created a repository through signed Nostr events - verified authorized pushes and rejected unauthorized clone/push - pushed a surviving `master` branch - deleted the active `main` branch over authenticated Smart HTTP - freshly cloned the repository and verified `master` became HEAD, `origin/main` was absent, and repository content remained intact Signed-off-by: Tal Weiss <major.tal@gmail.com> |
||
|
|
b7bb15122e |
feat(projects): add buzz projects CLI commands (NIP-MP kind:30621) (#4020)
Implements the `buzz projects` command group — the NIP-MP Phase 2 write path for kind:30621 multi-repo projects. The relay accepted kind:30621 in #3171; this adds the two-layer Rust builder in `buzz-sdk` and the seven CLI commands. ## What this adds ### `crates/buzz-sdk/src/builders.rs` — two-layer builder **Layer A (protocol):** - `validate_project_envelope(tags, content)` — 8 NIP-MP rules in relay order: `d`-cardinality, `d`-empty/length, member-cap (≤64 `a` tags, checked before per-tag parse), member-tag-arity (2–3 elements), member-coordinate grammar (first-two-colons split, literal `30617`, lowercase 64-hex owner, non-empty remainder), member-duplicate (coordinate only, hint ignored), singleton metadata cardinality, byte bounds (`name` ≤256 / `description` ≤2048 / `buzz-channel` ≤256 / `buzz-visibility` ≤256). - `build_project_with_tags(content, tags)` — raw Layer A builder; RMW mutations path. - `ProjectMemberCoord` — `30617:<owner-hex>:<repo-d>` + optional opaque relay hint; equality/Hash by coordinate only. **Layer B (writer policy):** - `build_project(slug, name, description, members, channel, visibility)` — constructs `d` tag, enforces UUID channel and `listed|unlisted` visibility, forces empty content; composes onto Layer A. This is the `create` path. **Shared:** - `build_delete_addressable(kind, pubkey, d)` — generic NIP-09 kind:5 coordinate delete; `build_workflow_delete` now delegates to this. - All 31 `NIP-MP.fixtures.json` cases exercised through `build_project_with_tags`; count assertion guards against omissions. ### `crates/buzz-cli/` — seven commands ``` buzz projects create <slug> --repo <coord> [--name] [--description] [--channel <uuid>] [--visibility listed|unlisted] buzz projects get <slug> [--owner <pubkey>] buzz projects list [--owner <pubkey>] [--limit <n>] buzz projects add-repo <slug> --repo <coord> [--repo <coord>]... buzz projects remove-repo <slug> --repo <coord> [--repo <coord>]... buzz projects update <slug> [--name|--clear-name] [--description|--clear-description] [--channel <uuid>|--clear-channel] [--visibility listed|unlisted|--clear-visibility] buzz projects delete <slug> ``` Command semantics: - **`create`**: all local validation (slug, repos, channel, visibility, name length) fires before the collision preflight — invalid input returns `Usage` without a network call. Routes through Layer B (`build_project`). - **`update`**: at least one setter/clearer required — enforced by a clap `ArgGroup` with `required(true).multiple(true)`, with a runtime backstop for programmatic callers; setter + own clearer are mutually exclusive per clap conflicts. - **`add-repo`/`remove-repo`**: coordinate expansion and dedup fire before head fetch — malformed or duplicate `--repo` values return `Usage` without touching the relay. - **`delete`**: head-based tombstone at `created_at = head + 1`; post-submit re-query verifies tombstone landed. - All mutations: strip `auth`, re-validate full envelope through Layer A; `created_at` advances from observed head, never wall-clock. - Relay hints on existing member tags preserved verbatim through RMW. ## Limitations (recorded, not in scope) - **No relay-hint authoring**: `--repo` carries a coordinate only; existing hinted `a` tags survive RMW unchanged. - **Signer-self delete only**: NIP-OA owner-delete extension not exposed; `delete` targets the signer's own coordinate. - **Deletion durability**: watermark carry-over applies; `delete` is best-effort against a later-arriving replacement. ## Live round-trip 21-step transcript executed against a relay built from `origin/main` `b1b283cd4`, covering create, get, multi-field update (name + description + channel in one call), channel set/clear, add-repo, remove-repo, delete (tombstone verified at `head+1`, repeated delete → `NotFound`). Delta transcript confirmed multi-field update, channel set/clear, no-op add-repo → `Conflict` exit 5, empty update and setter+own-clearer both rejected at parse time. Duplicate create → `Conflict`. Cross-owner `add-repo` with full coordinate exercised. --------- Signed-off-by: Will Pfleger <pfleger.will@gmail.com> Co-authored-by: npub1mn7jgtj4w2pd0g0zeuhxsa6jy6p0rewxz4kujt98my82ahfmp72sxjexk7 <dcfd242e557282d7a1e2cf2e6877522682f1e5c6156dc92ca7d90eaedd3b0f95@buzz.block.builderlab.xyz> |
||
|
|
89bf03c05d |
fix(nip-oa): accept raw Nostr tag form in parse_json_array (#4203)
## What `BUZZ_AUTH_TAG` stored in the **raw Nostr tag form** `[auth,hex,,hex]` (unquoted, comma-delimited — how an `auth` tag serializes inside a Nostr event and how `.env` files commonly store it) was rejected by the CLI: ``` BUZZ_AUTH_TAG is malformed: invalid JSON: expected value at line 1 column 2 ``` …and even when the CLI *could* parse it, it forwarded the raw string as the `x-auth-tag` header, so the relay's `verify_auth_tag` (which expects JSON) rejected it with `403 relay_membership_required`. Two commits close both gaps. ## Commits ### 1. `fix(nip-oa): accept raw Nostr tag form in parse_json_array` `parse_json_array` (`crates/buzz-sdk/src/nip_oa.rs`) only accepted well-formed JSON arrays. Added a fallback: when strict JSON parsing fails *and* the trimmed input is bracket-delimited, split on `,` and treat each field as a string (empty field `,,` → empty string, matching `["auth","hex","","hex"]`). All consumers (`parse_auth_tag`, `verify_auth_tag`, the CLI, `buzz-acp`) benefit from one change at the lowest layer. ### 2. `fix(cli): canonicalize BUZZ_AUTH_TAG to JSON before sending x-auth-tag header` The CLI stored the raw input string and sent it verbatim as the `x-auth-tag` header (`client.rs:618`). Added `canonicalize_auth_tag` in `buzz-sdk`: parse either form, re-serialize to canonical JSON. The CLI now canonicalizes before storing as `auth_tag_json`, so the header is always valid JSON regardless of input form. Together: local parse + wire canonicalization means the raw form works end-to-end. ## Why The raw form `[auth,hex,,hex]` is exactly how an `auth` tag serializes inside a Nostr event. That shape leaks into `.env` files and shell variables because there's no canonical "stored form" outside an event. The SDK + CLI should accept it rather than push quoting/conversion logic onto every consumer (harnesses, agent shells, external tools). ## Security Both changes are purely syntactic — they only change how a 4-element string array is extracted and containerized. All downstream validation is unchanged: - `parse_auth_tag`: still checks exactly 4 elements, `"auth"` label, 64-char lowercase-hex pubkey, 128-char signature. - `verify_auth_tag`: still reconstructs the preimage and verifies the BIP-340 Schnorr signature against the owner pubkey. No new attack surface — a malformed or forged tag is still rejected at the same validation points. ## Tests 4 new tests in `nip_oa::tests`: - `test_parse_auth_tag_raw_nostr_form` — raw form with conditions + empty conditions - `test_parse_auth_tag_raw_form_with_whitespace` — raw form with surrounding whitespace - `test_canonicalize_auth_tag_raw_to_json` — raw→JSON and JSON→JSON normalization All 25 `nip_oa` tests pass (21 existing + 4 new). `cargo fmt --check` and `cargo clippy -p buzz-sdk -p buzz-cli` clean. ## Verification Confirmed end-to-end against a live community relay (`wss://hermesagent.communities.buzz.xyz`): - **Before:** raw `BUZZ_AUTH_TAG` → CLI parse error, or `403 relay_membership_required` if somehow parsed. - **After:** raw `BUZZ_AUTH_TAG` → CLI parses it, canonicalizes to JSON for the header, relay accepts via NIP-OA owner delegation, `buzz channels members` returns the full roster. ## Context Originated from a community investigation where agent-side relay access was failing because the harness-exported `BUZZ_AUTH_TAG` (raw Nostr form) was rejected by the CLI (expecting JSON). This removes the impedance mismatch at the source. --------- Signed-off-by: amanning3390 <adam.manning@pro-serveinc.com> Signed-off-by: Tyler <109685178+tlongwell-block@users.noreply.github.com> Signed-off-by: npub1qyvc0c5kl4gqv2fd97fsk46tu378sqgy35vc83rvgfwne90sel7s0ed67d <011987e296fd5006292d2f930b574be47c7801048d1983c46c425d3c95f0cffd@buzz.block.builderlab.xyz> Co-authored-by: npub1qyvc0c5kl4gqv2fd97fsk46tu378sqgy35vc83rvgfwne90sel7s0ed67d <011987e296fd5006292d2f930b574be47c7801048d1983c46c425d3c95f0cffd@buzz.block.builderlab.xyz> Co-authored-by: Tyler <109685178+tlongwell-block@users.noreply.github.com> |
||
|
|
ac4fa13b8e |
perf(relay): serve relay-membership checks from the read replica (#4124)
## Summary Route `Db::is_relay_member` — the membership check that runs on every authenticated HTTP request and WS AUTH — through the standard `route_read` machinery on the bounded arm, instead of adding a bespoke cache (replaces #3844). - `crates/buzz-db/src/relay_members.rs`: add `is_relay_member_on(&mut PgConnection, ...)` executor seam; the pool version delegates to it. - `crates/buzz-db/src/lib.rs`: `Db::is_relay_member` now routes via `route_read("relay_membership", RoutePredicate::Bounded)` — replica only on a proved fresh session, writer on any route rejection, writer re-run on replica query error. Exactly the shape of every other routed read. This is the one permission read served from the replica, by explicit product decision (Tyler accepted ≤1s bounded staleness on reads we choose): the fleet-wide fence guarantee (`BUZZ_REPLICA_READ_MAX_AGE_MS`, deploy target 1s) is an order of magnitude tighter than the 10s TTL proposed in #3844 and needs no invalidation machinery. Staleness is symmetric for admits and revokes. `BUZZ_REPLICA_READ_MAX_AGE_MS` unset = writer-only = kill switch. It is not precedent for routing other permission reads. ## Validation At this exact commit (`git rev-parse HEAD` confirmed in the same shell, rustc 1.95): - `cargo test -p buzz-db` — 94 passed, 0 failed - PG-gated suite single-threaded — **151 passed, 2 failed**; the 2 failures are the per-owner-limit tests broken on main by #3829 (limit 3→5, tests still seed 3) — they fail identically at base `19d57b0d4` in a pristine control checkout; separate trivial fix to follow - New PG-gated test `is_relay_member_is_bounded_routed_and_fails_closed` — divergent writer/replica fixtures prove: budget unset ⇒ writer; budget set + fresh proof ⇒ replica; over-budget entry ⇒ writer - clippy `-D warnings` + fmt clean; pre-push hooks green (desktop check/test, rust tests, tauri checks) - **Live-local pass** (TESTING.md, release binary, `BUZZ_REQUIRE_RELAY_MEMBERSHIP=true`, fresh DB): - writer-only (no `READ_DATABASE_URL`): member accepted, outsider 403 `relay_membership_required`; metrics `route_decision{path="relay_membership",decision="writer",reason="disabled"}` - replica configured + `BUZZ_REPLICA_READ_MAX_AGE_MS=1000`: member accepted / outsider denied via `decision="replica",reason="fresh"`; admit visible to the routed check within ~1.2s; revoke enforced within ~1.2s - reader outage mid-flight (TCP proxy killed): member send still succeeds in <200ms via `decision="writer",reason="reader_acquire_timeout"`; outsider still denied — fails closed, no availability loss Reviewed by Wren: 9/10 minimalness, 9/10 elegance, 9.5/10 correctness at this SHA. Supersedes the 10s-cache approach in PR 3844, which should be closed unmerged once this lands. Signed-off-by: Tyler <109685178+tlongwell-block@users.noreply.github.com> Co-authored-by: npub1qyvc0c5kl4gqv2fd97fsk46tu378sqgy35vc83rvgfwne90sel7s0ed67d <011987e296fd5006292d2f930b574be47c7801048d1983c46c425d3c95f0cffd@buzz.block.builderlab.xyz> |
||
|
|
5765fc74b7 |
fix(relay): allow open relays to set their NIP-11 workspace icon (kind:9033) (#3998)
## Problem The desktop deliberately shows the workspace icon editor on open relays (#2640, gate: `canEditIcon` in `desktop/src/features/communities/ui/EditCommunityDialog.tsx`) and defers to the relay-side kind:9033 check — which required an admin/owner row in `relay_members`. For a community with **no admin/owner row at all** (the `ensure_configured_community` path, which never writes an owner), every 9033 was refused and the icon was permanently unsettable. **Correction from review (thanks @Dawn):** the original version of this PR claimed nobody holds a role on an open relay. That's false — `main.rs` bootstraps `RELAY_OWNER_PUBKEY` as owner regardless of `BUZZ_REQUIRE_RELAY_MEMBERSHIP`, so a production open relay like bb-block *does* have an owner row, and the old gate was refusing everyone except that owner. The first revision of this diff would have silently widened that owner-only control to any NIP-42-authenticated sender. ## Fix — steward-wins `may_set_workspace_profile(sender_role, membership_enforced, community_has_steward)`: | Relay mode | Community has admin/owner row? | Who may set the icon | |---|---|---| | Closed (`require_relay_membership=true`) | any | admin or owner (unchanged) | | Open | yes (e.g. bb-block) | admin or owner (unchanged posture) | | Open | no (genuinely rosterless) | any NIP-42-authenticated sender | - New DB helper `has_admin_or_owner(community)` (`crates/buzz-db/src/relay_members.rs`); the call site only queries it on open relays. - The rosterless admit logs a `warn!` with the sender pubkey — 9033 writes no audit row and publishes no announcement event (unlike 9030/9031), so this is the only durable attribution. - Kinds 9030–9032, NIP-42 auth, `AdminUsers` scope, ban gate, and icon validation are all untouched. - Doc comment fixed: cited nonexistent `canEditCommunityProfile`; real symbol is `canEditIcon`. ## Test coverage — closing the mutation gap Dawn's mutation testing showed the original unit tests pinned only the helper's truth table: inverting the flag at the call site or deleting the gate entirely survived the full suite. - Unit tests now cover the 3-arg truth table (closed steward-independent, open-with-steward stays steward-only, rosterless-open admits). - Two `#[ignore]`d Postgres integration tests drive `handle_relay_admin_event` with a real `AppState` (open rosterless admit → steward appears → roleless refused again; closed relay member refused). Wired into the Backend Integration CI job as a dedicated nextest step. - **Both of Dawn's mutants verified killed** at this head: flag inversion fails 1 unit test; gate deletion fails both integration tests (`Ok(())` where `Rejected` expected). ## CI wrinkle found and fixed: pre-existing schema drift The first Backend Integration run of the new 9033 tests failed with `column "icon" of relation "communities" does not exist` — migration `0003_community_icon.sql` added the column, but `schema/schema.sql` (the desired-state file that CI job applies via pgschema) was never updated. Pre-existing drift, invisible until a test in that job actually wrote the column. Fixed in `297148f62` (3-line addition to `schema/schema.sql`). ## Receipts (at `1b4b52db8` code / `297148f62` head) - `cargo test -p buzz-relay`: 835 pass, 1 fail — `api::mesh_demo::tests::demo_join_forwarded_arm_round_trips_echo`, pre-existing (fails identically at the old base and on clean main); `telemetry::trace_context_lookup_does_not_enable_callsites` is a known order-dependent flake, passes in isolation. - `cargo test -p buzz-db`: 94 pass. - Both ignored integration tests pass live against local Postgres. - `cargo fmt --all -- --check`: clean. - Live-local pass per TESTING.md at this head (release build, relay on :3199, real WS + NIP-42 via nak): - open rosterless: roleless key sets icon → NIP-11 serves it; `warn!` with sender pubkey in the relay log - open + owner row inserted: fresh roleless key refused ("must be admin or owner"); owner sets icon - closed relay (owner bootstrapped, `BUZZ_RELAY_PRIVATE_KEY` set): plain member refused, owner sets icon, `javascript:` URL rejected, empty icon clears (NIP-11 → null) --------- Signed-off-by: npub1qyvc0c5kl4gqv2fd97fsk46tu378sqgy35vc83rvgfwne90sel7s0ed67d <011987e296fd5006292d2f930b574be47c7801048d1983c46c425d3c95f0cffd@buzz.block.builderlab.xyz> Co-authored-by: npub1qyvc0c5kl4gqv2fd97fsk46tu378sqgy35vc83rvgfwne90sel7s0ed67d <011987e296fd5006292d2f930b574be47c7801048d1983c46c425d3c95f0cffd@buzz.block.builderlab.xyz> |
||
|
|
b1b283cd4c |
fix(buzz-acp): thread cache-read tokens into NIP-AM kind:44200 events (#3999)
## Problem `buzz-agent` measures and sends `accumulatedCachedInputTokens` on the wire (`usage.rs:93`). `buzz-acp` deserializes it correctly — but then drops it: `TurnUsage` had no cache field, and `build_turn_metric_counts` hardcoded `cache_read_tokens: None` and `cache_write_tokens: None` into both `turn` and `cumulative` `TokenCounts`. Every kind:44200 event published permanently lacked data the harness measured. The archive is append-only — this is unrecoverable data loss per turn, every turn, until fixed. NIP-AM already specifies the fields (`cacheReadTokens` / `cacheWriteTokens` inside `turn` and `cumulative`). This is a pure threading fix. ## Changes **`crates/buzz-acp/src/usage.rs`** - `SessionState` gains `last_cached_input: u64` to track the committed cache-read baseline. - `TurnUsage` gains `turn_cache_read_tokens: Option<u64>` (field-local; `None` when no baseline or counter decreased) and `cumulative_cache_read_tokens: u64` (always present; zero when no cache hits reported). - `record()` computes the cache-read delta with field-local taint semantics: a decrease in the cumulative counter nulls only `turn_cache_read_tokens` — it does not flip `delta_reliable` or invalidate `turn_input_tokens`/`turn_output_tokens`. Identical to the `accumulatedTotalTokens` pattern already present. - `take()` and the setup-notification branch both advance `last_cached_input` in the committed baseline. **`crates/buzz-acp/src/pool.rs`** - `build_turn_metric_counts` wires `turn_cache_read_tokens` into `turn.cache_read_tokens` (when `delta_reliable`) and `Some(cumulative_cache_read_tokens)` into `cumulative.cache_read_tokens`. - `cache_write_tokens` remains `None` on both counts with an explanatory comment: buzz-agent does not emit a write-side count on the wire today. - Six existing `TurnUsage` struct literals in tests updated with the two new fields. ## Tests **`usage.rs` — new cache-read section (5 tests):** - `cache_read_first_turn_produces_none_turn_delta_and_passes_cumulative_through` — no baseline → delta None, cumulative passes through - `cache_read_second_turn_delta_computed_correctly` — delta = current − previous - `cache_read_decrease_nulls_turn_cache_but_leaves_delta_reliable` — field-local taint: decrease nulls cache delta only, input/output stay reliable - `cache_read_zero_payload_after_baseline_produces_zero_delta` — zero on both sides → `Some(0)`, not `None` - `cache_read_threads_through_setup_notification_baseline` — setup notification baseline correctly seeds the cache counter **`pool.rs` — new acceptance test (1 test):** - `test_build_turn_metric_counts_cache_read_tokens_thread_through` — wire-parses a buzz-agent payload with nonzero `accumulatedCachedInputTokens`, runs two turns through the tracker and `build_turn_metric_counts`, and asserts nonzero `cacheReadTokens` in cumulative + correct per-turn delta in `turn`; also asserts `cache_write_tokens` is `None` throughout ## Quality gates at tip `c6405eb43f532572e3b7775e0dee826dc9cb3f82` | Gate | Result | |---|---| | `cargo test -p buzz-acp` | **655/655**, 0 failed | | `cargo clippy -p buzz-acp --all-targets -- -D warnings` | clean | | `cargo fmt --check` | clean | Note: the pre-push hook `mobile-test` gate fails on `origin/main` before this branch (Flutter test in `channels_page_test.dart` / `compose_bar_test.dart` — verified independently). My changes touch only `crates/buzz-acp/src/`; the mobile failure is unrelated and pre-existing. --------- Signed-off-by: Will Pfleger <pfleger.will@gmail.com> Co-authored-by: npub1g8493u0xfsjrvflg4n08ezd7vec99mnwzlv0qgwpr9d7gvjwhuzqx59rhw <41ea58f1e64c243627e8acde7c89be667052ee6e17d8f021c1195be4324ebf04@buzz.block.builderlab.xyz> |
||
|
|
cb9701cd30 |
feat(relay): accept kind:30621 multi-repo projects at ingest (#3171)
Buzz renders one card per `kind:30617`, so a project spanning several repositories has no representation. [NIP-MP](https://github.com/block/buzz/pull/3163) defines `kind:30621` as an addressable container holding a group's name, description, channel binding, and member coordinates. This adds the kind to `buzz-core` and its structural validation to the relay ingest path. ## Event shape ```json { "kind": 30621, "tags": [ ["d", "platform"], ["name", "Platform"], ["description", "Relay, desktop, and mobile."], ["a", "30617:<owner-a-hex>:buzz"], ["a", "30617:<owner-b-hex>:buzz-infra"], ["buzz-channel", "<channel-uuid>"], ["buzz-visibility", "listed"] ] } ``` ## Validation at ingest | Rule | Behavior | |------|----------| | `d` tag | exactly one, non-empty (length already bounded by the generic `D_TAG_MAX_LEN` check) | | member `a` tag arity | exactly 2 or 3 elements per NIP-01's `a` tag grammar; a 4th element has no defined meaning and is rejected | | member `a` tag coordinate | must parse as `30617:<lowercase-64-hex-owner>:<non-empty-d>` | | duplicate members | rejected on exact string match of the canonical coordinate | | member cap | 64, counted over raw `a` tags | | metadata cardinality | at most one each of `name`, `description`, `buzz-channel`, `buzz-visibility` | | metadata length | `name` ≤ 256 bytes, `description` ≤ 2048 bytes, `buzz-channel` ≤ 256 bytes, `buzz-visibility` ≤ 256 bytes | | zero members | valid | | unknown tags | ignored | Rejection order is normative so a client can predict which rule fires: `d`-cardinality → `d`-empty → member-cap → member-arity → coordinate parse → member-duplicate → metadata cardinality → metadata length. ## Design notes **No membership authorization.** Members are `a` tags, so one project may name repositories owned by different pubkeys — the entire point of the kind. That is safe because membership grants nothing: push policy reads a repository's own `kind:30617` (`api/git/policy.rs`) and never a project. `buzz-channel` is a metadata reference, not a routing directive, so projects are classified global-only. **Owner-only editing is free.** NIP-33 addressing keys replacement on `(pubkey, kind, d)`, so one signer can never overwrite another's project. No relay-side permission check exists or is needed, and `test_project_same_d_under_two_authors_are_independent` pins it. **Duplicates are rejected, not deduped.** A relay cannot rewrite tags inside a signed event without invalidating its id and signature, so the alternative to rejection is a stored duplicate-member head that every consumer must apply a first-wins rule to. **The cap is checked before the duplicate set is built.** Counting raw `a` tags rather than distinct coordinates means an event naming one coordinate thousands of times is refused on count, instead of being bounded only by the relay frame limit. **No side-effect handler.** Generic NIP-33 replacement and generic NIP-09 coordinate soft-delete already cover replacement and deletion; `kind:30621` needs no entry in `is_side_effect_kind`. ## Generic NIP-09 fix carried along `soft_delete_by_coordinate` (`crates/buzz-db/src/event.rs`) previously deleted the live coordinate head regardless of the tombstone's own `created_at`, so a delayed or replayed `a`-tag deletion signed between two versions destroyed the newer replacement. NIP-09 scopes an `a`-tag deletion to versions at or before the deletion request, so the `UPDATE` now carries `created_at <= $5` and `handle_a_tag_deletion` threads the deletion event's `created_at` through. The bug predates `kind:30621` and affected every parameterized-replaceable kind on the generic path — `kind:30617` repository announcements included — so the fix lands there rather than as a project special case. `events.created_at` is immutable per row, so the predicate guarantees a tombstone can never erase a version newer than itself; the UPDATE re-evaluates its WHERE clause after any lock wait. Under READ COMMITTED, a same-coordinate replacement racing the deletion may cause the deletion to evaluate before the new head lands, returning `Ok(false)` — but that outcome is state-identical to the deletion having arrived first, a valid Nostr ordering Nostr never fixes. The return value feeds only a debug log. No coordinate-level lock is needed. ## Coverage 32 unit tests in `crates/buzz-relay/src/handlers/ingest.rs` pin the envelope contract (accept: minimal, cross-owner, zero-member, same repo `d` under two owners, colon-bearing repo `d`, cap boundary, unknown tags, relay hint on member `a` tag, max-length metadata, stranger-owned member, uninterpreted metadata values, non-empty content; reject: every rule above plus valueless `d`/`a` tags). A fixture-driven test (`project_envelope_validates_all_shared_fixtures`) runs every case in the shared `NIP-MP.fixtures.json` oracle (11 accept + 20 reject) against `validate_project_envelope`, so any future change that breaks a case turns the test suite red. 6 `#[ignore]`d e2e tests in `crates/buzz-test-client/tests/e2e_project.rs` cover behavior that only exists past storage — coordinate round-trip, newer-wins replacement, two authors sharing a `d`, an `a`-tag tombstone that removes the project while leaving referenced `kind:30617`s intact, and a tombstone timestamped between V1 and V2 that must leave V2 live. The negative e2e case asserts on the rejection message so a refusal for an unrelated reason cannot satisfy it; that is what proves the validator is reachable from the live write path rather than merely correct in isolation. The new e2e binary is wired into the Relay E2E job. The timestamp predicate is additionally pinned at the storage layer by `coordinate_delete_spares_head_newer_than_the_deletion` in `crates/buzz-db/src/lib.rs`, which asserts both directions: a stale tombstone deletes nothing and leaves the newer head readable, and a tombstone at the head's own timestamp still deletes it. This test is wired into the Backend Integration job. Related: #3163 (the NIP-MP spec and shared conformance fixtures). Independent — either can merge first. --------- Signed-off-by: Will Pfleger <pfleger.will@gmail.com> Co-authored-by: npub1mn7jgtj4w2pd0g0zeuhxsa6jy6p0rewxz4kujt98my82ahfmp72sxjexk7 <dcfd242e557282d7a1e2cf2e6877522682f1e5c6156dc92ca7d90eaedd3b0f95@buzz.block.builderlab.xyz> |
||
|
|
c104eecfb3 |
feat(desktop): import local Pocket voices (#3259)
## Context Pocket TTS currently offers bundled reference voices. People also need a local, private way to add a voice without sending audio to a cloud service. ## Summary Add a Pocket voice import flow to Voice settings. Buzz opens the native file picker, decodes common audio formats in the reusable `buzz-voice` crate, canonicalizes the selected audio, stores it under a content-derived identity in app data, selects it, and lets the user delete it later. ## Changes - Accept WAV, M4A, MP3, FLAC, OGG, and AIFF files between 2 and 30 seconds, including multichannel sources. - Decode and downmix accepted audio to canonical mono 32 kHz PCM16 WAV before hashing and storage. - Store imported voices behind stable `pocket:imported:<sha256>` identities and content-addressed files. - Keep absolute file paths inside the native process and expose only voice metadata to React. - Include imported voices in Pocket preview and live huddle playback. - Add Add voice and delete controls while preserving the bundled Pocket voice catalog. - Fall back to Mary when the selected imported voice is deleted. - Keep durable import, selection, and deletion successful when a live TTS worker acknowledgement is delayed. - Preserve bundled voices when optional import metadata is unreadable and keep failed deletion retryable. ## Related issue None found. ## Testing Production decoding was exercised with WAV, M4A with AAC, MP3, FLAC, OGG Vorbis, and AIFF fixtures. Each format canonicalized to mono 32 kHz PCM16 WAV. Manual validation in the combined daily-driver build covered native-picker import, Preview, live-huddle playback, deletion, and Mary fallback. ## Screenshots The Voice settings card preserves the bundled Pocket catalog and adds the local Add voice action.  ## Reviewer-reproducible examples Create common-format fixtures and run them through the production importer: ```bash . ./bin/activate-hermit fixtures="$(mktemp -d)" ffmpeg -hide_banner -loglevel error -f lavfi -i "sine=frequency=220:duration=3" -ac 2 -ar 44100 "$fixtures/voice.wav" ffmpeg -hide_banner -loglevel error -i "$fixtures/voice.wav" -c:a aac "$fixtures/voice.m4a" ffmpeg -hide_banner -loglevel error -i "$fixtures/voice.wav" "$fixtures/voice.mp3" ffmpeg -hide_banner -loglevel error -i "$fixtures/voice.wav" "$fixtures/voice.flac" ffmpeg -hide_banner -loglevel error -i "$fixtures/voice.wav" -c:a libvorbis "$fixtures/voice.ogg" ffmpeg -hide_banner -loglevel error -i "$fixtures/voice.wav" -c:a pcm_s16be "$fixtures/voice.aiff" BUZZ_VOICE_IMPORT_TEST_DIR="$fixtures" \ cargo test -p buzz-voice imports_common_audio_format_fixtures -- --ignored --nocapture ``` Exercise import persistence, synthesis, deletion, and bundled-voice fallback with an installed Pocket model: ```bash BUZZ_POCKET_MODEL_DIR=/path/to/pocket-model-bundle \ cargo test -p buzz-voice --test pocket_import_audio \ objective_import_synthesis_delete_and_mary_fallback \ -- --ignored --nocapture ``` Exercise the native-picker boundary, selection, preview dispatch, deletion, cancellation, and invalid-file states: ```bash cd desktop pnpm build:e2e pnpm exec playwright test tests/e2e/voice-settings.spec.ts --project=smoke ``` --------- Signed-off-by: John Tennant <jtennant@block.xyz> Signed-off-by: John Tennant <johnmatthewtennant@gmail.com> Signed-off-by: John Tennant <jtennant@squareup.com> Signed-off-by: npub1qyvc0c5kl4gqv2fd97fsk46tu378sqgy35vc83rvgfwne90sel7s0ed67d <011987e296fd5006292d2f930b574be47c7801048d1983c46c425d3c95f0cffd@buzz.block.builderlab.xyz> Signed-off-by: npub12gtutshhh76rx0jx697f32f9tffd4hhp3hx58fp4x6u4uemkm7sqf8f757 <5217c5c2f7bfb4333e46d17c98a9255a52dadee18dcd43a43536b95e6776dfa0@buzz.block.builderlab.xyz> Co-authored-by: John Tennant <jtennant@block.xyz> Co-authored-by: npub1qyvc0c5kl4gqv2fd97fsk46tu378sqgy35vc83rvgfwne90sel7s0ed67d <011987e296fd5006292d2f930b574be47c7801048d1983c46c425d3c95f0cffd@buzz.block.builderlab.xyz> Co-authored-by: npub12gtutshhh76rx0jx697f32f9tffd4hhp3hx58fp4x6u4uemkm7sqf8f757 <5217c5c2f7bfb4333e46d17c98a9255a52dadee18dcd43a43536b95e6776dfa0@buzz.block.builderlab.xyz> |
||
|
|
61ba9dfaa0 |
refactor(voice): extract reusable Pocket primitives + Pocket voice settings (relands #2467 + #3208) (#3910)
Relands **#2467** (extract `buzz-voice` crate) and **#3208** (Pocket
voice settings) onto main, after #3266 and #3180 merged.
## Why a fresh PR
The repo is squash-only with delete-branch-on-merge. Squashing #3266
deleted `jtennant/pocket-tts-2026-04`, which was #2467's base — GitHub
auto-closed #2467 and it cannot be reopened. Squash merges also sever
ancestry, so GitHub's natural merge-base reports phantom conflicts for
the whole remaining stack.
## Content provenance
- Byte-identical to the blessed `jt/buzz-voice-refactor` branch
(`93029c577`, tree `6729e0eff` — reviewed by Dawn (#2467) and Max
(#3208) at exact heads) **except** the three files where #3180 and #3208
genuinely interact.
- Three-file resolution (union of both sides):
- `huddle/mod.rs` — #3180's pipeline re-exports + #3208's
`agent_tts_routing` imports.
- `huddle/state.rs` — `reset_preserving_generation` preserves both
`huddle_generation` (#3180) and `tts_enabled` (#3208); test sets merged
into one `tests` module.
- `desktop/src/testing/e2eBridge.ts` — both switch arms kept; no
duplicate case labels.
## Verification at
|
||
|
|
081f805d5e |
feat(agent): optional reply guard reminds a silent turn to publish (#3763)
## Why
A Buzz agent's assistant text and reasoning are never shown to anyone —
only what it posts through the CLI. A turn that runs fifteen tool calls
and never publishes is a silent failure: the requester waits on a result
that was produced and thrown away.
This adds an optional reminder at the end-of-turn gate, off by default.
Tyler asked for it in buzz-mesh; plan iterated to **9.5/10 with @Wren**
(Minimalness 9.7, Elegance 9.5, Correctness 9.3).
## What
`BUZZ_AGENT_REQUIRE_REPLY=1` (default off, per-agent opt-in). A turn
about to end with no recognized attempt to post gets a reminder and is
rerolled. **At most two, then the turn ends regardless** — the guard
catches accidental omission, it does not compel speech. The reminder
text explicitly licenses silence so it cannot fight the base prompt's
"silence is usually correct."
**This is not a new MCP hook.** `RunCtx::run` *is* the turn, so the two
per-turn locals need no plumbing, and every tool call already passes
through it with arguments visible. The objection is appended at the
existing `_Stop` gate and rides `push_hook_outputs_as_tool_results`, so
the model receives it as a lower-trust tool result with `{hook, server,
text}` attribution. No new trust path, no new lifecycle event, no
dev-mcp or CLI protocol change.
Earlier revisions of this plan needed four crates (a `_UserPromptSubmit`
hook, a marker file, a `buzz-cli` change, dev-mcp state). Tyler pointed
out the agent already knows both facts; that deleted all of it. Net
runtime change is ~35 lines in `agent.rs` + ~4 in `config.rs`.
### Recognition contract
A registered non-hook tool whose qualified name ends in `__shell`, whose
`command` argument contains `messages send` or `reactions add`.
- **The `__` separator is exact, not approximate.** Given `has()` +
`!is_hook()`, `ends_with("__shell")` is *provably equivalent* to a bare
name of `shell`: registration forbids `__` in server and bare names
(`mcp.rs:227,268`) and qnames are `{server}__{bare}`, so a trailing
`__shell` could only straddle the separator if the bare name began with
`_` — which `is_hook` excludes. Without the separator, `powershell` and
`noshell` would match.
- **Reads the structured `command` field**, not serialized arguments, so
a `description` that quotes a send cannot disarm the guard, and a
non-string `command` is rejected rather than coerced.
- **Detects an attempt, not a successful publish.** A failed send
already returns non-zero exit and error JSON — louder than this
reminder. The variable is named `buzz_reply_call_seen` so the code can't
pretend otherwise.
- **Checked after the per-turn tool-call cap**, since a discarded call
never ran.
- `messages send` also covers `messages send-diff`. Reactions count
because the base prompt directs agents to react rather than post a bare
acknowledgement.
**Known limits, both deliberate and documented:** a command assembled at
runtime (`$CMD`) or hidden in a wrapper script is missed; text that
merely quotes a send (`echo "buzz messages send"`) matches. Missing a
real post is the expensive direction and substring matching is the
forgiving one there. Neither edge is pinned by a test, so the matcher
stays free to improve.
### Budget
Reminders share `BUZZ_AGENT_STOP_MAX_REJECTIONS`, the existing outer cap
on every end-turn objection. Default 3 fits both; at 1 only one fits; at
0 the guard is off with the hooks. A round carrying both a hook
objection and a reminder costs one rejection and delivers both texts. An
independent budget would either violate that bound or need a second
arbitration rule.
## Prior art
- **#3467** (closed) built the same detector one layer up in `buzz-acp`
for a different remedy. None of its symbols are on main — this borrows
its permission to be coarse, but reads structured data that ACP didn't
have.
- **#3648** (open) detects turns with *no output at all*; a turn with
fifteen tool calls and no post counts as output there, so it does not
cover this case.
- **#3741** (merged) is mesh-only.
## Testing
**14 new tests.** 4 unit tests on the matcher; 10 integration tests
through the ACP wire harness: off by default, `=0` still off, opted-in
silent → exactly 2 reminders then `end_turn`, registered `fake__shell`
send → 0 reminders, hallucinated `fake__shell` → still reminded, publish
call truncated past the 64-call cap → still reminded, budget 1 → 1
reminder, budget 0 → off, combined `_Stop` hook objection + reminder →
one round both texts and after 2 reminders the hook objection continues
alone, unparseable `=true` → startup error naming the key.
**10 mutation checks, each breaking a specific named test** — neutralize
the nag cap, stop sharing the budget, neutralize `buzz_reply_call_seen`,
drop `has`/`is_hook`, ignore the flag, drop the `__`, drop `reactions
add`, read serialized args, move detection before truncation.
`tests/bin/fake_mcp.rs` gains `FAKE_MCP_SHELL_TOOL=1`: it previously
exposed no tool with a bare name of `shell`, so the satisfied-guard path
was untestable.
Full `cargo test -p buzz-agent` green at 9e0ae1f04; clippy `-D warnings`
and `cargo fmt --check` clean.
**Unrelated flake found:**
`cancelled_turn_with_usage_emits_notification_before_response`
(`tests/fake_llm.rs`) is timing-sensitive. Under 10 loaded cores it
fails **2/20 on this branch and 1/20 at unmodified
`origin/main@02be413b8`** — pre-existing, not caused by this change
(which is inert without the env var). Flagging so it isn't misattributed
to the next PR that's open when CI hits it.
## Docs
`crates/buzz-agent/README.md` is the primary home (env var, recognition
contract, limits, budget interaction). `docs/MCP_DRIVEN_HOOKS.md` gets a
short cross-reference explaining this is *not* a hook — otherwise
readers hunt for a `_ReplyGuard` tool that doesn't exist.
---------
Signed-off-by: tlongwell-block <109685178+tlongwell-block@users.noreply.github.com>
Signed-off-by: npub1mprnacetjua2xx3p5eddmhxyk6wv929ymm5py8kd2xfxurxahspqqlgyta <d8473ee32b973aa31a21a65adddcc4b69cc2a8a4dee8121ecd51926e0cddbc02@buzz.block.builderlab.xyz>
Co-authored-by: Dawn (sprout agent) <c6237ef84fa537c78dcee78efd2d4e59f728859c7f194da42ac51ededfa0be05@sprout-oss.stage.blox.sqprod.co>
Co-authored-by: npub1mprnacetjua2xx3p5eddmhxyk6wv929ymm5py8kd2xfxurxahspqqlgyta <d8473ee32b973aa31a21a65adddcc4b69cc2a8a4dee8121ecd51926e0cddbc02@buzz.block.builderlab.xyz>
|
||
|
|
10d5a26414 |
feat(relay): raise hosted community limit to five (#3829)
## Summary - raise the relay authoritative default community ownership limit from 3 to 5 - raise the desktop hosted-community treatment from 3 to 5 - preserve `BUZZ_MAX_COMMUNITIES_PER_OWNER` as a deployment override ## Validation - `pnpm -r check` - `cargo fmt --all -- --check` - `cargo test -p buzz-db` (94 passed, 151 Postgres-dependent tests ignored) - pre-push hooks: desktop checks/tests, Rust tests, Tauri checks (all passed; 1,995 desktop Rust tests passed) Signed-off-by: npub1dccv64krpcpse5cmkzfeh998cftungyatw3djt8jwdw6g43f7fyqzzmrf7 <6e30cd56c30e030cd31bb0939b94a7c257c9a09d5ba2d92cf2735da45629f248@buzz.block.builderlab.xyz> Co-authored-by: npub1dccv64krpcpse5cmkzfeh998cftungyatw3djt8jwdw6g43f7fyqzzmrf7 <6e30cd56c30e030cd31bb0939b94a7c257c9a09d5ba2d92cf2735da45629f248@buzz.block.builderlab.xyz> |
||
|
|
23f0c26b1c |
fix(relay): align NIP-11 max_limit with REQ ceiling (#3635)
Buzz's NIP-11 document advertised `limitation.max_limit: 10_000`, but the effective websocket REQ page ceiling was `1_000` — a 10x lie. The websocket REQ path never sets `EventQuery::max_limit`, so `query_events` applied its own `unwrap_or(1000)` clamp to every historical query. Only the COUNT fallback (`apply_count_fallback_limit`) ever raises that clamp. A client that trusts the advertised value asks for 10,000 events, silently receives 1,000, and — with no error and no continuation signal — reads that short page as exhaustion. Up to 9,000 events are dropped without anyone noticing. `MAX_HISTORICAL_LIMIT = 2_000` in `handlers/req.rs` was dead weight for the same reason: nothing clamped to 2,000 could survive the DB's 1,000 clamp one layer down. ## Change `buzz_db::DEFAULT_MAX_PAGE_LIMIT` (`1_000`) is now the single source of truth. It is the `query_events` clamp default, the value both REQ clamp sites use, and the value advertised as NIP-11 `max_limit`. `MAX_HISTORICAL_LIMIT` is removed rather than re-pointed — an alias for a constant used four lines away adds a name without adding meaning. The NIP-50 search path carries a second, independent bound. It clamps its emission target to the shared ceiling like any other REQ, but how many FTS candidates it will scan was bounded separately, by a bare 10-page loop over 100-hit pages. That product only coincidentally equalled the ceiling, so raising the ceiling — or shrinking a page — would shrink the scan relative to what clients may now request, degrading search quality while nothing in the code registered the change. The page count is now ceiling-divided from `DEFAULT_MAX_PAGE_LIMIT` over a named `SEARCH_PAGE_SIZE`, so the scan budget tracks the advertised ceiling by construction. That budget is a resource policy, not a delivery promise. It bounds candidates *scanned*, not events *emitted*: post-filtering (NIP-01 match, channel access, reader visibility, dedup) discards an unpredictable share of every page, so a search result smaller than the requested limit remains possible. This is not a NIP-11 violation — `max_limit` is defined as a clamp the relay applies to a requested `limit`, not a guaranteed count in the response. Two guards hold the pair together: - `req_filter_limit_clamps_to_advertised_nip11_max_limit` reads `max_limit` back out of a built `RelayInfo` and asserts the REQ path clamps to exactly that number. - `search_scan_capacity_covers_advertised_nip11_max_limit` asserts the scan budget covers exactly one advertised ceiling's worth of candidates — no less, and with no spare page of slack, so the derivation can't be quietly replaced by a hand-tuned constant that happens to pass today. ## Behavior Websocket behavior is unchanged: 1,000 was already the real ceiling on every path, including NIP-50. The advertisement now tells the truth about it. Raising the effective limit is a capacity decision and is deliberately not made here. The generic HTTP bridge's page-2+ offsets do change, as a consequence of the corrected clamp. `extract_page_offset` sizes a page from `query.limit` *before* the DB clamp applies, so an absent limit previously produced an offset of 2,000 and a requested 1,500 produced 1,500 — while the page actually returned held at most 1,000 rows. Both now produce 1,000. This corrects paging that had been skipping rows the previous page never returned; `extract_page_offset_sizes_pages_from_clamped_limit` locks it down. ## Scope note The bridge's per-endpoint ceilings — `BRIDGE_WINDOW_MAX_LIMIT` (200) for channel windows and `BRIDGE_THREAD_MAX_LIMIT` (500) for thread reads — are endpoint contracts on a non-NIP-01 transport, not values NIP-11 speaks for, and are unchanged. Fixes #3757 --------- Signed-off-by: Will Pfleger <pfleger.will@gmail.com> Co-authored-by: Duncan <dcfd242e557282d7a1e2cf2e6877522682f1e5c6156dc92ca7d90eaedd3b0f95@buzz.block.builderlab.xyz> |
||
|
Will PflegerandGitHub
|
114d40d9d3 |
feat(relay): gate kind 30178 team-catalog reads behind the shared tag (#3358)
Team catalog projections (`kind:30178`) embed every member's system
prompt, so they need the same read gate personas already have: only the
author sees an unshared event. The gate was hardcoded to `kind:30175` at
six read surfaces plus the SQL pushdown, so rather than adding a second
special case it becomes kind-generic over `SHARED_GATED_KINDS = {30175,
30178}`.
## Kind 30178
New parameterized-replaceable kind, addressed by `(pubkey_o, 30178,
team_id)`. It embeds sanitized member projections instead of referencing
`kind:30175` heads — a foreign reader of a shared team could not
otherwise hydrate members whose own persona events are unshared or, for
built-ins, absent entirely. `kind:30176`'s wire body is untouched, so
device sync keeps its contract.
## Kind-generic shared gate
`buzz_core::kind` replaces `is_persona_shared_kind` /
`is_unshared_persona_event` / `persona_event_is_shared` with
`SHARED_GATED_KINDS` and the kind-agnostic `is_shared_gated_kind` /
`is_unshared_gated_event` / `event_is_shared`. Every read surface
consults the set:
| Surface | File |
|---|---|
| REQ historical delivery + `ids` lookup |
`crates/buzz-relay/src/handlers/req.rs` |
| Live fan-out | `crates/buzz-relay/src/handlers/event.rs` |
| COUNT fallback | `crates/buzz-relay/src/handlers/count.rs` |
| NIP-98 HTTP `/query`, `/count`, `/search` |
`crates/buzz-relay/src/api/bridge.rs` |
| Pre-`LIMIT` SQL pushdown | `crates/buzz-db/src/event.rs` |
The SQL clause generalizes from `kind != 30175` to `kind NOT IN (...)`
bound from `SHARED_GATED_KINDS`, still applied before `ORDER BY … LIMIT`
so a page of newer private events cannot starve an older shared one off
the candidate set. `EventQuery::persona_reader` is renamed
`shared_gated_reader` and `needs_persona_filtering` to
`needs_shared_gate_filtering` to match.
Because the `buzz-core` rename has consumers outside the relay, the four
desktop call sites of `persona_event_is_shared` travel with it:
`desktop/src-tauri/src/commands/personas/pending.rs`,
`desktop/src-tauri/src/event_sync.rs`, and two in
`desktop/src-tauri/src/managed_agents/persona_events.rs`. Each call is
unchanged apart from the name — the persona `shared` projection behaves
exactly as before.
## Ingest validation
`validate_persona_envelope` splits into two reusable pieces —
`validate_shared_tag` (exactly-two-element `["shared","true"]`, at most
one occurrence) and `single_bounded_d_tag` (exactly one `d` tag,
non-empty, `<=64` chars, no ASCII control characters or whitespace).
`validate_team_catalog_envelope` composes both; personas additionally
keep the slug grammar `^[a-z0-9][a-z0-9_-]{0,63}$`.
`kind:30178` deliberately does **not** get the slug grammar. Team ids
are UUIDs or built-in identifiers such as `builtin-team:welcome`, and
the colon is not slug-legal; rewriting ids to fit would break NIP-33
addressing against the team's own `kind:30176` head. The non-empty and
exactly-one checks are load-bearing regardless — without them generic
NIP-33 storage maps a missing `d` onto `(pubkey_o, 30178, "")` and every
team overwrites its predecessor.
The exact two-element `shared` shape is enforced because the SQL
visibility clause is JSONB containment (`tags @>
'[["shared","true"]]'`), which would match a three-element superset such
as `["shared","true","extra"]`.
`kind:30178` is also added to the `Scope::UsersWrite` allowlist and to
`is_global_only_kind`, so a stray `h` tag cannot channel-scope an
owner-authored definition.
## Deferred
`kind:30176` is deliberately not a gate member. Its writers never emit
`shared`, so catalog opt-in semantics do not describe it — it needs
owner-private reads driven by an authenticated principal set, tracked as
a separate follow-up.
## Tests
- 19 new `ingest.rs` unit tests covering the 30178 envelope (UUID and
colon `d` tags, 64-char boundary, non-ASCII bound,
empty/valueless/duplicate/missing `d`, embedded newline, `shared`
false/three-element/duplicate, scope and global-only membership).
- Persona regressions for the valueless `["d"]` shapes, since the
`d`-tag helper is shared by both validators.
- Existing `kind.rs` gate tests generalized and extended to assert the
gate applies to 30178 as it does to 30175.
- New `crates/buzz-test-client/tests/e2e_team_catalog.rs`: 9 WS-level
tests over a live relay covering author reads of unshared heads, foreign
omission from REQ, `ids`-lookup denial, COUNT existence-leak, share and
unshare transitions, and the mixed-kind filter case.
- `.github/workflows/ci.yml` adds `--test e2e_team_catalog` to the Relay
E2E job so the new suite runs.
## Docs
`docs/nips/NIP-AP.md` gains a "Team catalog projection: kind:30178"
section and an "Ingest validation: kind:30178" subsection, records the
gate as kind-generic, documents 30178 deletion vs. unshare semantics,
and adds a security note that sharing a team exposes every member's
instructions even when that member's own `kind:30175` head is unshared.
Signed-off-by: Will Pfleger <pfleger.will@gmail.com>
|
||
|
MatthewandGitHub
|
dba97eecd9 |
fix(db): isolate usage metrics advisory-lock test on scratch DB (#3670)
## What Problem This Solves `test_usage_metrics_lock_has_single_owner_and_releases_on_drop` hardcodes the **production** advisory lock key (`0x4255_5A5A_4D45_5452`) on the shared `TEST_DATABASE_URL`. Postgres advisory locks are per-database, so any live `buzz-relay` pointed at the same DB holds that key and the test fails (or races the relay tick). Diagnosis time was burned during #3268 verification, including near-misses on live dev relays. Fixes #3619. ## Why This Change Was Made Preferred fix from the issue: run the test on a private scratch DB via existing `create_scratch_db` / `drop_scratch_db` (same pattern as replica-routing fixtures). Keep the production lock key so the test still documents the real constant, without colliding with a running relay. ## User Impact - Local `cargo test -p buzz-db -- --ignored` no longer fails when a dev relay is running against the shared test DB - Safer: no temptation to `pg_terminate_backend` a live relay to "fix" the test ## Evidence - Code review of fixture isolation - Pattern matches existing `create_scratch_db` usage in this file - Test remains `#[ignore = "requires Postgres"]` (same as before) ## Related - Issue: #3619 - None found among open PRs for this exact fix Signed-off-by: NanoRisk6 <aidashtherapy@gmail.com> |
||
|
|
bf139e8d0b |
perf(presence): reduce heartbeat frequency (#3783)
## Summary - send desktop presence heartbeats every 60 seconds instead of every 30 seconds - extend presence TTL from 90 to 180 seconds to preserve the existing three-heartbeat expiry window - add mutation-sensitive tests that pin the one-minute / three-window timing contract - update presence documentation to match This halves steady-state **desktop** presence `SET` + `PUBLISH` traffic while retaining tolerance for two missed heartbeats. Mobile already uses a 60-second heartbeat, so the fleet-wide reduction depends on desktop's share of connected clients. ## Rollout order Deploy the relay TTL increase before shipping the desktop heartbeat change. Old desktop + new relay is safe; new desktop + old relay leaves only a 90-second TTL on a 60-second cadence and can flap after one missed heartbeat. ## Verification At initial live-test commit `00816e233b187bc5ba12c667d675ed050a8cc1c9`: - isolated clean-room relay built from the exact SHA against fresh Postgres, Redis, and MinIO - live Redis `MONITOR` observed kind-20001 writes as `SET ... EX 180`, global `PUBLISH`, and clean-disconnect / explicit-offline `DEL` - normal workflows passed: channel create/update/archive/unarchive; message send/get/reply/thread/search; archived-channel write rejection and resumed write after unarchive At follow-up commit `bf38a8c5c96f196ff8ee46e48d4141ee7811f186`: - `pnpm -C desktop test` — 3829 passed - `pnpm -C desktop typecheck` - `cargo test -p buzz-pubsub` — 24 passed, 11 Redis-dependent tests ignored - mutation probes fail when the server TTL changes to `999999` or the desktop heartbeat changes back to 30 seconds - `git diff --check` The pre-push suite's relevant checks passed, but its unrelated Tauri clippy step fails on current `origin/main`: `desktop/src-tauri/src/linux_media.rs` has three dead-code warnings on macOS. This PR does not modify that file, so the branch was pushed after independently running the suites above. ## Buzz context Originating channel: `buzz-redis-cluster-mode` (`f4e36d32-afdb-447f-8c87-ab003e069d18`) --------- Signed-off-by: npub1t2tgm7d8f995uqvmnm8h88sg3wnpp9a5xysjf6dg3tjmgt3ltulqdp8ehr <5a968df9a7494b4e019b9ecf739e088ba61097b4312124e9a88ae5b42e3f5f3e@buzz.block.builderlab.xyz> Co-authored-by: npub1t2tgm7d8f995uqvmnm8h88sg3wnpp9a5xysjf6dg3tjmgt3ltulqdp8ehr <5a968df9a7494b4e019b9ecf739e088ba61097b4312124e9a88ae5b42e3f5f3e@buzz.block.builderlab.xyz> |
||
|
|
53771c8f54 |
fix(acp): preserve truncated thread context (#3340)
## Why Long Buzz threads were rendered as `[Thread Context (13 of 13 messages)]` because the harness counted only the already-limited query result. That hid older context and could also hide the agent's own prior reply in busy threads. ## What - Fetch one extra thread reply as a sentinel so truncated context is labeled correctly. - Use a best-effort `/count` call for improved truncated totals when available, clamped to the sentinel-proven minimum so racy counts cannot render impossible labels. - Keep the `/count` path single-attempt with a short timeout and only add the root to exact totals when the root was actually fetched. - Fetch and preserve the agent's newest prior reply when it falls outside the recent window, with exact event-id matching for the pin/dedup boundary. - Add parser and fetch-boundary tests for truncation, exact count, missing root, count-below-minimum clamping, count failure fallback, distinct fetched-reply lower bounds, agent-reply dedup/pinning, and serialized query/count filter semantics. ## Risk Assessment Low-to-medium — limited to buzz-acp prompt context fetching and a small RestClient helper. If `/count` fails or times out, the code falls back to the sentinel-derived minimum total rather than failing the prompt. The synchronous `/count` happens only for truncated thread contexts and is bounded to one short best-effort attempt. ## References - Buzz thread: chotchkies-buzz-bombing-flakes / `7ef71407f1c7a642382c7e48e0c80fb6ca66948890e04d1eb6f1408c3b7278b1` - Validation at `c1cfd1b16a04a3ac1d1d0d3cf43e1a08508f3532`: - `cargo fmt -p buzz-acp` ✅ - `cargo test -p buzz-acp test_fetch_thread_context -- --nocapture` ✅ (6 tests) - `cargo test -p buzz-acp parse_nostr_thread_response` ✅ - `cargo test -p buzz-acp` ✅ (649 unit + 9 lifecycle tests) - `git diff --check` ✅ - Push was completed with `--no-verify` after pre-push hooks reached non-code local environment failures: `flutter` missing for `mobile-test`; Node.js v20.20.2 too old for pnpm/node:sqlite in `desktop-check` and `desktop-test`. Earlier hook stages passed: `check-push-org`, `branch-skew`, `rust-tests`, `test`, `desktop-tauri-checks`. - Earlier full `./bin/just ci` at `622ed7eb8807d64e06209101569b1013414af091` ⚠️ passed Rust/desktop/web stages, then failed in `mobile-test` on unrelated existing mobile test `ChannelDetailPage keeps follow mode off while a tall newest message stays visible`; rerunning that single mobile test reproduced the same failure without touching mobile code. Generated with Codex Signed-off-by: npub1m0vvn9qm5md0a080p27qzkm9uaw49e699ukwfq7fc0756xq0y5zqhzhdk2 <dbd8c9941ba6dafebcef0abc015b65e75d52e7452f2ce483c9c3fd4d180f2504@buzz.block.builderlab.xyz> Co-authored-by: npub1m0vvn9qm5md0a080p27qzkm9uaw49e699ukwfq7fc0756xq0y5zqhzhdk2 <dbd8c9941ba6dafebcef0abc015b65e75d52e7452f2ce483c9c3fd4d180f2504@buzz.block.builderlab.xyz> |
||
|
|
4933672eb4 |
feat(mesh): upgrade embedded mesh to v0.74 and harden shared compute (split 1/2 of #3467) (#3741)
## Summary This is **part 1 of 2** split out from #3467 (per Tyler's request), carrying only the mesh-scoped changes. The agent/ACP response-behavior changes and the new `send_message` tool stay in #3467 as part 2. All commits are @michaelneale's work, cherry-picked with authorship preserved. - Upgrade embedded Mesh to v0.74.0 (tag-pinned instead of commit rev) and use canonical Gemma model IDs. - Keep shared compute serving through member joins, roster changes, app recovery, and community switching. - Wait for actual model readiness and avoid resuming incomplete downloads after quit. - Leave `BUZZ_AGENT_THINKING_EFFORT` unset by default so each model's chat template picks its own thinking default (`none` suppressed Gemma tool-calling entirely; pinning `low` made Qwen3 burn ~4x output budget). Explicit agent/persona/global values still win. ## Relationship to #3467 Contains the mesh commits from #3467 (`2cd640b23`, `0ad81c341`, `ad13ed841`) rebased onto current main, with one deliberate exclusion: the `crates/buzz-agent/src/llm.rs` reasoning→text parser change from `2cd640b23` is **not** here. That change unconditionally affects every OpenAI-compat/Responses provider, so it belongs with the reply-behavior work in part 2, where it can be reviewed as what it is. Not included (remaining in #3467 / part 2): - typed `send_message` tool in dev-mcp + `BUZZ_ACP_SEND_MESSAGE_TOOL` gating - plain-reply delivery fallback in buzz-acp (`BUZZ_ACP_DELIVER_PLAIN_REPLIES`) - the mesh_agent_e2e P5/P6 rewrite (exists to prove the reply path) - the two `env.insert` preset opt-ins in `relay_mesh.rs` for the flags above - the llm.rs parser change This PR is independently mergeable; part 2's flags are all off by default so it can land before or after. ## Testing - `cargo test -p buzz-relay --locked` — 780 passed (one telemetry test is order-sensitive under parallel default settings; passes in the pre-push suite and standalone, unrelated to this diff — files untouched here). - `just desktop-tauri-test` (default features) — 1877 passed. - `cargo test --locked --features mesh-llm` in `desktop/src-tauri` — 1961 passed, including the new relay-mesh preset and coordinator/recovery tests. - Both `Cargo.lock`s resolve with `--locked` against the v0.74.0 tag. - Full pre-push hook suite green (rust-tests, desktop-check/test, tauri checks). Live validation of the mesh v0.74 upgrade itself is documented on #3467 (two-Mac cross-version test). --------- Signed-off-by: Michael Neale <michael.neale@gmail.com> Signed-off-by: Tyler Longwell <tlongwell@block.xyz> Co-authored-by: Michael Neale <michael.neale@gmail.com> Co-authored-by: Tyler Longwell <tlongwell@block.xyz> |
||
|
|
262f2392e3 |
fix(cli): resolve agents from owner records (#3178)
## Context
`buzz users get --name Honey` searches relay-wide profiles and can
return an identically named agent owned by someone else. This caused
agents from the wrong owner to be added to a channel.
## Summary
This bug fix scopes exact-name agent lookup to owner-authored
managed-agent records, then cryptographically verifies each returned
profile's NIP-OA `auth` tag before asserting ownership. The relay and
database contracts remain unchanged.
### Related issue
None found.
## Changes
- Adds `buzz users get --name Honey --owner me|<hex>|<npub>`.
- Resolves `me` to the NIP-OA owner when the CLI runs as an agent,
otherwise to the CLI identity.
- Matches kind `30177` managed-agent record names exactly and
case-insensitively under the requested owner.
- Requires exactly one valid NIP-OA `auth` tag whose verified owner
equals the requested owner and whose `kind` and `created_at` conditions
apply to the profile event before returning `owner_pubkey` or
`owned_by_me: true`.
- Keeps missing, malformed, stale, condition-mismatched, or unverifiable
owner-record candidates visible with `owned_by_me: false` and an
explicit `verification` value.
- Returns every same-name record for the owner so callers can require
explicit selection when duplicates remain.
- Preserves the existing output shape and client-side name filter for
unscoped searches.
- Documents the distinct owner-scoped managed-agent lookup and unscoped
NIP-50 lookup modes.
### Testing
The reviewer-reproducible red and green commands below exercise the
ownership bug against the target branch and this branch.
## Screenshots
Not applicable. This is a CLI-only change.
## Reviewer-reproducible examples
The lookups below were run against the live relay from `main` and this
branch.
### Red: unscoped lookup returns the 100-profile relay-wide cap and
excludes John's agents
On `main`:
```bash
cargo run -q -p buzz-cli -- users get --name Honey \
| jq '{count: length, first_three: .[:3] | map(.pubkey), johns_agents: map(select(.pubkey == "31b29bcbe69d6716fbb7ba33602b89200bfc9ddfdabcfd1ea6fbfa70b816dfc7" or .pubkey == "4597ac725bba33fc7dd0454c1e2316a5ed770426acf667837d46f6553b3fcf54"))}'
```
Observed output:
```json
{
"count": 100,
"first_three": [
"20d27fc6c0ab4f50b66d1a32a64c5ca1fb985254143ce911f61ab7733333c3d7",
"00644478cdd9032c563ddc712b3687d8345d948945aab3c18bab95afbf6f519a",
"93c16697d0e58007bc11fb953208bc6b1cff387b2dee094abc10bf82dfee5424"
],
"johns_agents": []
}
```
`main` also rejects the owner-scoped command:
```bash
cargo run -q -p buzz-cli -- users get --name Honey --owner me
```
```text
error: unexpected argument '--owner' found
Usage: buzz users get --name <NAME>
```
### Green: owner-scoped lookup distinguishes verified and unresolved
records
On this branch:
```bash
cargo run -q -p buzz-cli -- users get --name Honey --owner me \
| jq 'map({pubkey,display_name,owner_pubkey,owned_by_me,verification})'
```
Observed output:
```json
[
{
"pubkey": "0ca77314d7ac8b3fcf6c647cc8cb9c3afd840db3b2a8ff2079f09a168de1827e",
"display_name": null,
"owner_pubkey": null,
"owned_by_me": false,
"verification": "missing_profile"
},
{
"pubkey": "31b29bcbe69d6716fbb7ba33602b89200bfc9ddfdabcfd1ea6fbfa70b816dfc7",
"display_name": "Honey",
"owner_pubkey": "67252b09c31a995daa63aada26569fbc6a3d12f573113f001ce7432f870da820",
"owned_by_me": true,
"verification": "verified"
},
{
"pubkey": "4597ac725bba33fc7dd0454c1e2316a5ed770426acf667837d46f6553b3fcf54",
"display_name": "Honey",
"owner_pubkey": "67252b09c31a995daa63aada26569fbc6a3d12f573113f001ce7432f870da820",
"owned_by_me": true,
"verification": "verified"
}
]
```
Only the two profiles with valid NIP-OA proofs assert ownership. The
owner-authored record whose profile is absent remains visible but cannot
be selected as verified ownership.
---------
Signed-off-by: npub1qye6rec0htgg3np8yt6plpyyg8cyffaq66emt3kmk05eylckkzhq0hnf2k <0133a1e70fbad088cc2722f41f848441f044a7a0d6b3b5c6dbb3e9927f16b0ae@buzz.block.builderlab.xyz>
Co-authored-by: npub1qye6rec0htgg3np8yt6plpyyg8cyffaq66emt3kmk05eylckkzhq0hnf2k <0133a1e70fbad088cc2722f41f848441f044a7a0d6b3b5c6dbb3e9927f16b0ae@buzz.block.builderlab.xyz>
|
||
|
|
63496cc1d4 |
feat(replica): portable heartbeat-token fence with snapshot-local reader routing (#3268)
## Summary
Expands read-replica usage on the relay per the Rev 2 design (thread
39d1e174 in #buzz-read-only-replica-usage): replaces the
Aurora-incompatible WAL-LSN read-side fence observation with a portable
**heartbeat token**, makes the freshness proof **snapshot-local to the
serving reader session**, and adds a default-off bounded-staleness gate
for head fetches.
### Probe (writer side, unchanged ordering)
The ordered writer scan is kept verbatim — `S = clock_timestamp()` →
masked-visibility `pg_stat_activity` oldest-xact scan — and now ends by
committing a heartbeat token **last** on the same pinned connection
(single-row `UPDATE replica_heartbeat ... RETURNING token, epoch`,
migration `0026`). The single-row update serializes all pods' probes, so
tokens are globally commit-ordered; the three-bucket completeness
argument carries over unchanged. Ring retains `(token, committed_at,
fence_wall)`; epoch change resets the ring; a same-epoch token
regression (restore adversary) clears the ring **and rotates the epoch
on the writer** so pre-rewind readers fail the epoch check. Cadence 1s.
### Routing (snapshot-local proof)
Every routed read opens `BEGIN ISOLATION LEVEL REPEATABLE READ, READ
ONLY` on a reader session and observes the heartbeat as the
transaction's **first statement** — the proof's snapshot is exactly the
snapshot the page, participants batch, and bridge aux closure read from
(`ReadSession` carries the open transaction; drop = rollback).
Fail-closed everywhere: begin failure, missing heartbeat row, epoch
mismatch, token below the ring, over-budget entry → writer.
- **Predicate B (cursor pages, default-on):** same completeness math as
the existing fence — cursor timestamp must be ≤ the proved wall; thread
candidate-terminal and above-wall pages re-run on the writer.
- **Predicate A (head fetches, default-OFF):** gated by
`BUZZ_REPLICA_HEAD_MAX_AGE_SECS` (0 = off, clamped to fence staleness).
Bounded-stale head semantics are an explicit product decision — **do not
enable anywhere without Tyler's backdated-event-semantics acceptance.**
### Observability
`buzz_db_route_decision{path, decision, reason}` across all five paths,
`buzz_db_replica_heartbeat_age_seconds` gauge, and per-decision debug
logs carrying the proved token plus backend identity: `addr:port pid=N`,
prefixed with `aurora_db_instance_identifier()` when the endpoint
supports it (probed once per process on an autocommit checkout; SQLSTATE
42883 caches a definitive false; identity is evidence, never a routing
gate).
## Review & verification
- **Wren:** full review 9/9/9 at `5f81b10e5`; identity delta re-review
approved at exact head `fedb46368` (90/90 unit, clippy `-D warnings`,
fmt independently reproduced).
- **Max:** local E2E **PASS** at `5f81b10e5` — isolated PG17 writer +
two streaming standbys behind HAProxy; paused-reader legs split exactly
as designed (6 replica/fresh + 6 writer/stale), recovery clean, RR
snapshot hardening observed in runtime logs. Evidence:
`WORK_LOGS/2026-07-28_REPLICA_HEARTBEAT_REPLACEMENT_SHA_E2E.md`.
- **My gates at head (same shell):** buzz-db 90 unit + 143
Postgres-gated green (scratch DBs); buzz-relay 767/768 — lone red is the
pre-existing `mesh_demo` flake, red on base; clippy `-D warnings` + fmt
clean.
- Key regression tests:
`routed_request_holds_one_snapshot_across_page_and_aux`
(mutation-verified both directions), same-epoch rotation,
capability-probe negative, head-gate truth table, divergent-fixture
routing suite.
## Post-merge plan
Merge publishes the immutable `sha-*` main image → bb-public PR pins it
→ ArgoCD sync → Max runs the production cursor-routing canary on Aurora
(positive identity branch proven live). Head gate stays off. Client
`since`-widening ships separately.
---
## Update — full read routing (Rev 6, commit `9fa3c9c0b`)
Extends routing to the remaining read seams per
`PLANS/REPLICA_FULL_READ_ROUTING_DESIGN.md` Rev 6 (thread cf5deba4 in
#buzz-read-only-replica-usage).
### New routed seams — ALL deploy-default dark
The new seams (`query_events_routed`, `query_events_routed_bounded`,
`count_events_routed`, `get_events_by_ids_routed`,
`query_feed_{mentions,needs_action,activity}_routed`) are
**Bounded-only** and gated on `BUZZ_REPLICA_READ_MAX_AGE_MS`: unset ⇒
every new seam records `writer/disabled` and merging is a no-op. COUNT
and feed/by-ids never take the covered arm (deletion visibility: covered
bounds insert-completeness only). **Note:** the pre-existing cursor
paths (channel windows, thread pages) are *not* gated by this env var —
they route at B=0 today and that status quo is unchanged.
### Reader pool (D4/D5)
- Lazy pool (`connect_lazy`, `min_connections(0)`): reader-down at boot
can't crash the relay; a warn-only boot ping is the only boot-time
visibility, and it primes the Aurora identity capability cache. Priming
is an optimization only — the routed path itself spends a single acquire
budget regardless (see the single-checkout fix below), because a boot
ping that *fails* is correlated with exactly the reader-unavailable case
the budget bound exists for.
- `READER_ACQUIRE_TIMEOUT` = 150ms; a miss fails closed to the writer
with reason **`reader_acquire_timeout`** — named for the mechanism, not
a diagnosis. The budget includes cold connects and sqlx's `size` counts
in-flight dials, so this metric alone does not distinguish contention
from slow connection establishment (see `proved_reader` doc-comment for
the runbook guidance; the pool gauges are 10s samples and are for
capacity planning).
- `BUZZ_DB_READ_POOL_SIZE` sizes the reader independently (invalid/0
inherits writer sizing); `read_pool_stats().max` reports the reader's
own ceiling.
### Community isolation (formal-model question)
Isolation is structural — an explicit `community_id = $n` predicate
compiled into every query builder; no RLS, no session-GUC tenant state
(zero `CREATE POLICY` across migrations). The `_on` variants reuse the
exact same builders with only the executor swapped. Proven by a
seven-seam two-community divergent-fixture test
(`routed_reads_are_confined_to_the_requested_community`),
mutation-tested by Dawn: all four single-predicate stubs killed; the
mention-join feeds are defended in depth (three independent predicates)
so only complete removal leaks there.
### Accepted limitation
D6: client-side staleness on bounded reads (up to `_MS`) is accepted
product behavior when the gate is enabled; gate stays off at merge.
## Update — single acquire budget per routed read (commit `dd26caa9f`)
Max found (and Dawn independently reproduced, 302–330ms measured) that
the boot-unavailable cold path spent **two** stacked
`READER_ACQUIRE_TIMEOUT` budgets: the Aurora capability probe did its
own `pool.acquire()` before `begin_with` acquired again. Fixed by
acquiring **once** per routed read — the capability probe runs on the
held connection (`reader_aurora_capability_on`) and the read-only
`REPEATABLE READ` transaction begins on that same connection
(`Transaction::begin` accepts a `PoolConnection` at `'static` via sqlx's
`MaybePoolConnection`). Reason codes unchanged; capability still never
negatively cached. Measured routed fallback: ~150ms (one budget).
Ships with a PG-gated regression fixture
(`routed_fallback_spends_one_acquire_budget_when_aurora_cache_is_cold`,
authored by Dawn): size-1 reader saturated, capability cache asserted
cold, routes through `count_events_routed`, asserts writer answer +
one-budget elapsed + `writer/reader_acquire_timeout` label
(mutation-tested: `pool_busy` and `reader_validation_error` mutants both
killed). It fails at 330ms on the previous commit and passes on this
one.
### Verification at `dd26caa9f` (pinned toolchain rustc 1.95.0 via repo
`bin/`; ambient rustc 1.89 fails sqlx resolution — always prepend
`bin/`)
- buzz-db: 94 unit + **150/151** PG-gated serial (`--test-threads=1`;
suite is not parallel-safe against one Postgres). The 1 red is
`test_usage_metrics_lock_has_single_owner_and_releases_on_drop` — a
**pre-existing** same-key-same-database advisory-lock collision with any
live relay pointed at the shared `buzz` DB (the relay re-arms
`USAGE_METRICS_LOCK_KEY` on a 300s tick). It passes on a private scratch
DB with the relay still running — verified this run. Remedy is a scratch
`TEST_DATABASE_URL`, **not** terminating lock holders: inspect
`pg_locks`/`pg_stat_activity` and resolve the owning process first; an
idle advisory holder may be a live dev relay. Pre-existing, not a #3268
blocker — filed as #3619.
- buzz-relay: 773/773; clippy `-D warnings` + fmt clean. (`mesh_demo`
echo test is a known pre-existing Redis-dependent flake — Dawn confirmed
it fails identically on the base commit, untouched by this PR.)
- Dawn: independent re-verification at `dd26caa9f` — byte-identical diff
application confirmed, suite reproduced, and fixture proven to still
discriminate (fails at 320ms with only the production fix reverted).
Gate clear. Max: independent rig pass at `dd26caa9f` — 94/94 unit,
**15/15 routing/fallback matrix** (cursor behavior, default-dark seams,
seven-seam confinement, one-snapshot, dead-reader/no-URL fallback,
hard-delete fail-closed, reader max), 773/773 relay, private-DB lock
control. Gate clear; recommends merge + staged bb-block rollout
(no-reader-URL no-op → cursor-only → observe `buzz_db_route_decision` →
conservative nonzero `BUZZ_REPLICA_READ_MAX_AGE_MS`; rollback is
config-only).
## Known gap — CI does not run the PG-backed fixtures (#3622)
Dawn found post-sign-off (confirmed by Max and Eva at `dd26caa9f`):
**every Postgres-backed fixture in this PR is `#[ignore]`d and no CI job
selects it.** `Unit Tests` runs `cargo nextest run -p buzz-db --lib`
(`Justfile:279-285`, ignored tests excluded); the only `--run-ignored
ignored-only` invocations (`ci.yml:690`, `:702`) filter to
`relay_invite::tests`. So the one-budget regression, the seven-seam
isolation fixture, and the fence/floor-guard/fallback tests have zero
automated execution — the verification record above is exhaustive but
manual, at this exact SHA.
Not introduced by this PR (the `#[ignore]` + narrow-filter pattern
predates it; 34 ignored tests total) and not a merge blocker: all new
seams are dark until `BUZZ_REPLICA_READ_MAX_AGE_MS` is set. But it
changes the rollout gate — **do not set a nonzero
`BUZZ_REPLICA_READ_MAX_AGE_MS` in bb-block until CI enforces these
fixtures.** Filed as #3622 (explicit CI selection against the existing
backend-integration Postgres archive; ordering note: #3619 must land
first or be excluded, since widening the filter would select the
colliding usage-metrics lock test).
## Live redteam results (Max, real Helm + PG17 streaming replication at
`dd26caa9f`) — #3643, #3644
Max deployed this exact SHA via the repo chart against a physical
writer/standby pair and attacked it. **The PR's fenced routing passed
every fault**: healthy baseline routes proven on the real standby;
paused WAL replay moved head/cursor reads to `writer/stale`; B=0 moved
routed reads to `writer/disabled`; reader-URL removal rolled back
cleanly.
The redteam also found a **pre-existing** production risk this PR does
not create and does not fix: NIP-50 search builds its own eager,
unfenced pool straight from `READ_DATABASE_URL` (`main.rs:389-402`,
introduced by #2084, present in the deployed bb-block `sha-dd222a5` and
bb-public `sha-22be8bb` images, and both production ESOs already
template `READ_DATABASE_URL`). Measured: ~30s search stall on reader
loss, acknowledged-but-unsearchable writes under replica lag (unaffected
by `BUZZ_REPLICA_READ_MAX_AGE_MS`), fatal eager connect blocking pod
startup when the reader is down, and `replica=true` FTS even at B=0.
Filed as **#3643** (fix: pin FTS to the writer, or fence it like the
routed seams) with **#3644** for the writer-pointing-reader-URL
telemetry gap. Aurora's cluster-ro writer-fallback DNS softens the
outage modes in prod but not the lag mode.
**Revised ladder consequence:** step 2 ("add `READ_DATABASE_URL`, budget
unset") is NOT a no-op with today's binary — it hands FTS an unfenced
replica pool. Rollout order is now: merge (still a true no-op —
bb-block/bb-public already run the unfenced-FTS code and this PR only
adds dark seams) → fix #3643 → CI enforcement #3622 → then reader URL +
budget per the original ladder, re-running Max's three deployment tests
(reader-down latency, reader-down cold boot, paused-replay
read-your-own-write search).
## Activation gate (consolidated) — merge is clear; ALL of the below
precede any nonzero `BUZZ_REPLICA_READ_MAX_AGE_MS`
Six independent verification passes at `dd26caa9f` (Dawn ×2 incl.
byte-level + defang check, Max matrix + live Helm redteam, Wren full
CRUD regression + greenfield 0026 + 151/151 PG, Eva). Every induced
fault was either handled correctly by this PR's machinery or traced to
pre-existing main code. Remaining work gates **activation, not merge**:
1. **#3643** — unfenced NIP-50 FTS pool (pre-existing #2084; live in
prod today; also explains the blackhole search hang Wren observed — the
routed path was measured bounded at 151-152ms against a silent endpoint,
sqlx `inner.rs:252-255`).
2. **#3651** — reader `statement_timeout`: statements after acquire are
unbounded; mid-transaction blackhole reproduced hanging ≥15s (Dawn).
Small fix via `.after_connect` on the reader pool.
3. **#3622** — CI enforcement of the PG fixtures (with #3619 ordering).
4. **Recovery-conflict cancellation live test** — standby cancels a
routed read mid-snapshot → complete writer page, never partial (#3651's
fix also bounds the cancellation-never-arrives shape).
5. **DDL replication lag test** — migration on writer + paused replay +
budget on → fallback, not client-visible SQL errors.
6. **Reader-tx soak** — 30-60min mixed load; no idle-in-tx accumulation
on the standby.
Then Max's three deployment tests (reader-down latency, reader-down cold
boot, paused-replay read-your-own-write search) re-run on the fixed
binary before the first nonzero budget on bb-block.
---------
Signed-off-by: Tyler Longwell <tlongwell@block.xyz>
Signed-off-by: npub1qyvc0c5kl4gqv2fd97fsk46tu378sqgy35vc83rvgfwne90sel7s0ed67d <011987e296fd5006292d2f930b574be47c7801048d1983c46c425d3c95f0cffd@buzz.block.builderlab.xyz>
Co-authored-by: npub1qyvc0c5kl4gqv2fd97fsk46tu378sqgy35vc83rvgfwne90sel7s0ed67d <011987e296fd5006292d2f930b574be47c7801048d1983c46c425d3c95f0cffd@buzz.block.builderlab.xyz>
Co-authored-by: Tyler Longwell <tlongwell@block.xyz>
|
||
|
|
788b3c002b |
fix(git): channel binding tooling + author remediation for unbound repos (#3626)
Closes #3527. Repos announced via vanilla NIP-34 (kind:30617 without a `buzz-channel` tag) 404 forever: the SEC-005 read gate requires a channel-membership ACL, and nothing tells the author why or how to fix it. Per the ruling in the originating thread, this ships **bind/rebind tooling plus a narrow author-only remediation carve-out** — the shelved owner-circle approach is intentionally absent. ## Relay - **`api/git/binding.rs` (new):** shared tri-state binding resolver — `Bound(uuid)` / `NotBound` / `Broken`. First-tag, fail-closed: a malformed `buzz-channel` tag is `Broken`, never conflated with "no tag". Both gates use it. - **Read gate (`transport.rs`):** a **never-bound** repo read by **its own announcement author** still returns 404 (status byte-identical to the generic denial) but the body carries remediation: `run: buzz repos bind --id <repo> --channel <channel-uuid> — …`. This leaks nothing — the author announced the repo, and only the author can rebind (30617 is keyed by `(author, d)`). `Broken` bindings stay generic-denial for everyone, including the author (revocation shape). Bound-to-nonexistent-channel stays generic (phase 1; ingest validation is phase 2). - **Push gate (`policy.rs`):** unbound denial now returns `GIT_NO_CHANNEL_BINDING_BODY`. A deploy-skew test pins that the body carries both the new token (`no_channel_binding`) and the legacy phrase (`"no channel binding"`) so already-shipped desktops keep matching. **(Review r1, blocker 2)** `Broken` no longer collapses into "unbound": it denies 403 `invalid channel binding` for *everyone — including the announcement owner —* **before** the owner short-circuit, matching the read gate's fail-closed posture. The remediation token stays NotBound-only. - **`ingest.rs`:** side-effect failure `warn!` → `error!` — prod runs `RUST_LOG=error`, so these failures were invisible during triage. ## Contract - **`buzz-core/git_perms.rs`:** `GIT_NO_CHANNEL_BINDING_TOKEN` / `GIT_NO_CHANNEL_BINDING_BODY` consts as the declared cross-component contract; relay tests and desktop matcher both build on them. ## CLI - **`buzz repos bind --id <repo> --channel <uuid>`** — rebinds an existing announcement, preserving other tags. - **(Review r1, blocker 1)** **`--channel` on `buzz repos create`** — optional; injects exactly one shape-validated `buzz-channel` tag at creation via a pure `build_create_announcement` builder, so the primary create command stops producing repos the relay 404s. UUID existence/membership stays the relay's authority at git-access time (same TOCTOU posture as `repos bind`). Overlaps with #3594 (open, head 6bbe38459) — happy to reconcile whichever lands first; this branch also carries the bind path and tag preservation. ## Desktop - **Rust:** new `commands/project_git_merge_error.rs` (extracted from `project_git_workflow.rs` to respect the 1000-line ratchet); maps the token to a structured `no_channel_binding` error carrying the bind command. - **TS:** new `features/projects/lib/projectBranchErrors.ts` + tests — dual matcher (new token AND legacy spaced phrase); `ProjectBranchDialogs.tsx` uses it. ## Tests / verification (at head |
||
|
|
7012d86d52 |
feat: configure S3 URL addressing style (#3400)
## Summary - add one strict `BUZZ_S3_ADDRESSING_STYLE=path|virtual` setting shared by media and Git/CAS storage - preserve path-style defaults for bundled Compose/Helm MinIO while supporting Railway's virtual-hosted bucket contract - fail startup on invalid or non-Unicode values before dependency connection, and validate the Helm value with the same two choices - document operator mappings and why endpoint and bucket remain separate for routing and SigV4 signing ## Best-practice rationale AWS documents both URL forms and favors virtual-hosted addressing for S3, while compatibility endpoints such as the bundled MinIO deployment can require path style. `rust-s3` defaults to virtual/subdomain addressing and provides `with_path_style()` for the explicit compatibility case. Some providers buckets only support as virtual-hosted bucket styles. This PR therefore uses one explicit, provider-neutral switch rather than endpoint heuristics or fallback behavior, while retaining `path` as Buzz's backward-compatible default. Sources: - https://docs.aws.amazon.com/AmazonS3/latest/userguide/VirtualHosting.html - https://docs.rs/rust-s3/0.37.0/s3/bucket/struct.Bucket.html - https://docs.railway.com/storage-buckets#url-style - https://github.com/minio/minio/blob/master/docs/config/README.md#domain ## Validation - `cargo fmt --all` - `cargo check --workspace --all-targets` - targeted `buzz-media` and `buzz-relay` parsing/client-construction tests for defaults, strict errors, and both URL styles - Helm unittest: 45/45 passed - Compose config/render validation passed - local MinIO path-mode relay startup passed the Git A3 conformance probe and became ready - unreachable object storage failed startup and readiness never opened - push hooks completed the broader Rust and desktop suites successfully --------- Signed-off-by: npub122y0pqkertljmedu303rl0aqrj3w8pvu43t6jxm6875lzg6f2pwqegc3xc <5288f082d91aff2de5bc8be23fbfa01ca2e3859cac57a91b7a3fa9f12349505c@buzz.block.builderlab.xyz> Signed-off-by: Kalvin Chau <kalvin@block.xyz> Co-authored-by: npub122y0pqkertljmedu303rl0aqrj3w8pvu43t6jxm6875lzg6f2pwqegc3xc <5288f082d91aff2de5bc8be23fbfa01ca2e3859cac57a91b7a3fa9f12349505c@buzz.block.builderlab.xyz> |
||
|
|
ab55fee818 |
feat: add first-class OpenRouter provider support (#1975)
## Summary
First-class `Provider::OpenRouter` support joining the existing
anthropic/openai/databricks providers. Reuses the Chat Completions path
with targeted mutations for OpenRouter's routing contract.
**Core (`crates/buzz-agent`):**
- `Provider::OpenRouter` enum variant with `OPENROUTER_API_KEY`,
`BUZZ_AGENT_MODEL` → `OPENROUTER_MODEL` fallback, `OPENROUTER_BASE_URL`
env convention
- Body mutator: `reasoning: {effort}` when effort is configured, and
`max_completion_tokens` translated to OpenRouter's `max_tokens`
spelling; no `provider.require_parameters` filter (it routes only to
endpoints advertising every parameter in the body, which hard-404s a
valid model id); summaries get neither. `openai_body` is always called
with `effort=None` on the OpenRouter path — the `reasoning` object is
added by the mutator directly, so `reasoning_effort` is structurally
absent.
- Attribution headers: `HTTP-Referer: https://github.com/block/buzz`,
`X-OpenRouter-Title: Buzz`
- Error-inside-200 check in shared `parse_openai` (`finish_reason ==
"error"`)
- 401 auth handling: static API keys (`refresh_now` returns the same
token) fail terminal immediately with one wire request; PKCE/minting
sources get one retry with the fresh token.
- Status+`error_type` retry matrix (4-arm collapsed form): 429 (honor
`Retry-After`), 502 (retry), 503/`provider_overloaded` (honor
`Retry-After`), everything else including untyped 503 (bounded retries →
actionable routing message). 499 included matching shared `post()`
(#2175) for turn-timeout stall surfacing. Terminal failures wrapped in
`terminal_llm_error` for duration+attempt-count context.
- `anthropic/*` `cache_control` injection (model-gated, mixed-content
safe)
- Provider-agnostic `reasoning_details` opaque round-trip on
`HistoryItem::Assistant` for tool-call continuations — captured verbatim
in `parse_openai_with_reasoning_details`, replayed verbatim in
`openai_body`, byte-accounting charged. `provider_extra` passthrough
from `make_tool_call` composes independently.
**Desktop:**
- Readiness arms checking `OPENROUTER_API_KEY` + `OPENROUTER_MODEL`
- Model discovery via `{OPENROUTER_BASE_URL}/models` filtered on
`supported_parameters` contains `tools`
- Picker entry, credential config, effort table 3-file sync
**`desktop/src/features/agents/AGENTS.md`: no rules changed** — the
scoped rule requiring an explicit note is satisfied here.
Implements the gate-cleared plan from
`PLANS/OPENROUTER_PROVIDER_PLAN.md` (rev 3).
Signed-off-by: Will Pfleger <pfleger.will@gmail.com>
Co-authored-by: npub1mn7jgtj4w2pd0g0zeuhxsa6jy6p0rewxz4kujt98my82ahfmp72sxjexk7 <dcfd242e557282d7a1e2cf2e6877522682f1e5c6156dc92ca7d90eaedd3b0f95@buzz.block.builderlab.xyz>
|
||
|
|
f95fdc1a10 |
feat(agent,acp): wire provider total_tokens through NIP-AM publish chain (#3593)
## What Wires genuine provider-reported `total_tokens` through the full buzz-agent → buzz-acp publish chain so kind-44200 events carry real per-turn and cumulative totals for OpenAI-backed models, while preserving all existing behaviour for Anthropic and external harnesses (goose, claude-code). ## Why Live prod data showed 0 of 1,934 archived reports carry `totalTokens`. Both hardcoded `total_tokens: None` in `pool.rs` and the absent field in `buzz-agent`'s parser are root causes. This is the backend half of a two-track fix; the display-fallback half lands in [#2035](https://github.com/block/buzz/pull/2035). ## Changes **`crates/buzz-agent/src/types.rs`** - Added `total_tokens: Option<u64>` to `LlmResponse` with an explicit doc comment that NIP-AM forbids deriving it. - Added `TurnTotalState` enum (`Unseen | Exact(u64) | Unknown`) with `fold()` and `exact_value()` — the tri-state accumulator that distinguishes not-yet-observed from permanently poisoned. **`crates/buzz-agent/src/llm.rs`** - `parse_responses` and `parse_openai`: read `usage.total_tokens` from OpenAI Chat Completions (including Databricks routes) and the Responses API via `sum_usage`. - Anthropic: explicit `total_tokens: None` — no genuine total available; NIP-AM forbids summing categories. **`crates/buzz-agent/src/agent.rs`** - Added `turn_total_state: &'a mut TurnTotalState` to `RunCtx`. - Fold `response.total_tokens` into the accumulator after each usage-bearing response; non-usage-bearing responses (keepalive/stream frames) do not poison. **`crates/buzz-agent/src/lib.rs`** - Added `accumulated_total_state: TurnTotalState` to `Session` (default `Unseen`). - Per-turn state passed to `RunCtx`, folded into session cumulative after each turn. - Emits `accumulatedTotalTokens` in `usage_update` only when cumulative is `Exact(n)`. **`crates/buzz-acp/src/usage.rs`** - Added `accumulated_total_tokens: Option<u64>` (serde default) to `UsageUpdatePayload` — optional for goose compat. - Added `last_total: Option<u64>` to `SessionState`. - Added `turn_total_tokens` and `cumulative_total_tokens` to `TurnUsage` (field-local — never affect `delta_reliable`). - Derive turn-total delta only when prev and current are both `Some` and monotonic; absence, decrease, or no baseline leaves only the total delta null without touching input/output reliability. **`crates/buzz-acp/src/pool.rs`** - Replaced both hardcoded `total_tokens: None` in `publish_agent_turn_metric` with `usage.turn_total_tokens` and `usage.cumulative_total_tokens`. ## Tests 20 new tests across the four touched files: | File | Tests | |------|-------| | `types.rs` | `TurnTotalState` fold, accumulation, exact_value, default (7 tests) | | `llm.rs` | Chat present/absent, Responses present/absent, Anthropic always-None (5 tests) | | `usage.rs` | First turn no baseline, second-turn delta, cumulative decrease (field-local), current absent, goose-shaped deserialization, baseline absent (6 tests) | | `pool.rs` | Exact turn+cumulative mapping, null totals never derived (2 tests) | `cargo test -p buzz-acp -p buzz-agent` — all passing, 0 failures. ## Scope Boundary: `crates/buzz-agent/**` + `crates/buzz-acp/**` only. Desktop unchanged. `costUsd` explicitly out of scope. Related: [#2035](https://github.com/block/buzz/pull/2035) --------- Signed-off-by: Will Pfleger <pfleger.will@gmail.com> Co-authored-by: npub1g8493u0xfsjrvflg4n08ezd7vec99mnwzlv0qgwpr9d7gvjwhuzqx59rhw <41ea58f1e64c243627e8acde7c89be667052ee6e17d8f021c1195be4324ebf04@buzz.block.builderlab.xyz> |
||
|
|
005b5b819a |
feat(tracing): correlate trace IDs in relay logs (#3608)
## Summary Correlates trace + span IDs with logs, allowing traces and logs to be bridged seamlessly ### Related issue none found ### Testing Unit tests Signed-off-by: David Grochowski <dgrochowski@squareup.com> Co-authored-by: Amp <amp@ampcode.com> |
||
|
|
7adc46268d |
feat(cli): mirror Desktop mention delivery (#3330)
🤖 ## Summary Agent-authored mentions currently depend on matching visible `@Name` text to channel profiles. That makes notification delivery ambiguous when names collide or profiles change, and it encourages an extra post-send lookup just to confirm that the intended `p` tags were emitted. This change makes `buzz messages send` mirror Desktop's existing model: the message keeps a readable name in its content while the recipient pubkey is supplied separately. ```bash buzz messages send \ --channel <UUID> \ --content '@Alice could you review this?' \ --mention <alice-hex-or-npub> ``` `--mention` is repeatable. The CLI normalizes and deduplicates explicit pubkeys, merges them with any names it can resolve from the channel, and gives explicit identities priority under the existing 50-mention limit. Before uploading attachments, signing, or publishing, the command checks every resulting pubkey against the channel's current membership: - Members are mentioned normally. - Non-members stop the send and produce an actionable error. - `--allow-non-member-mentions` deliberately sends notifying `p` tags without adding anyone to the channel. Sending a message never changes membership. On success, `mention_pubkeys` is read from the exact signed event and returned with the relay response, so callers can verify the emitted recipients without another query. Managed-agent guidance teaches this single-command mention flow. Desktop mention behavior and the Nostr event schema are unchanged. Forum guidance is intentionally handled separately in #3596. ### Related issue None found. This replaces the earlier guidance-only approach in this PR with the underlying CLI behavior it required. ### Testing - `cargo test -p buzz-sdk` - `cargo test -p buzz-cli` - `cargo test -p buzz-acp` - `cargo test --manifest-path desktop/src-tauri/Cargo.toml` --------- Signed-off-by: npub1fdupjvyregj3z2tx7gx5x6py04zw89jm5usef9lyea4f3vcgh8qq9zgkdz <4b78193083ca25112966f20d4368247d44e3965ba7219497e4cf6a98b308b9c0@buzz.block.builderlab.xyz> Signed-off-by: npub13n66s06epmqf2kc3v373ez8hj65cuzyvxzjf93vwpervxqn2u7jq2qd9je <8cf5a83f590ec0955b11647d1c88f796a98e088c30a492c58e0e46c3026ae7a4@buzz.block.builderlab.xyz> Co-authored-by: npub1fdupjvyregj3z2tx7gx5x6py04zw89jm5usef9lyea4f3vcgh8qq9zgkdz <4b78193083ca25112966f20d4368247d44e3965ba7219497e4cf6a98b308b9c0@buzz.block.builderlab.xyz> Co-authored-by: npub13n66s06epmqf2kc3v373ez8hj65cuzyvxzjf93vwpervxqn2u7jq2qd9je <8cf5a83f590ec0955b11647d1c88f796a98e088c30a492c58e0e46c3026ae7a4@buzz.block.builderlab.xyz> |
||
|
Will PflegerandGitHub
|
ddd468723a |
revert(acp): remove dead GOOSE_ACP_SCHEDULER_DISABLED env injection (#3576)
## Summary [block/buzz#3144](https://github.com/block/buzz/pull/3144) injected `GOOSE_ACP_SCHEDULER_DISABLED=true` into every `AcpClient::spawn` call as a forward-compatible no-op, intended to suppress the cron scheduler in goose ACP children once the matching reader landed in goose. That reader only ever existed in [aaif-goose/goose#10738](https://github.com/aaif-goose/goose/pull/10738), which was closed unmerged. [goose#10781](https://github.com/aaif-goose/goose/pull/10781) (Lifei Zhou, merged 2026-07-29) disables the ACP scheduler by default at the source: `goose acp` now requires `--enable-scheduler` to start a scheduler. Buzz-spawned children therefore get no scheduler with zero configuration — making the `GOOSE_ACP_SCHEDULER_DISABLED` injection permanently dead code. ## What changes Removes from `crates/buzz-acp/src/acp.rs`: - `GOOSE_SCHEDULER_DISABLED_ENV` constant - `cmd.env(GOOSE_SCHEDULER_DISABLED_ENV, "true")` injection in `AcpClient::spawn` - `spawn_injects_scheduler_disabled_env_by_default` test - `spawn_scheduler_disabled_env_overrides_conflicting_extra_env` test - `spawn_and_read_child_env` helper (unreferenced once the two tests above are gone) No other files are affected. ## Why now Leaving dead code that references an env var no reader will ever consume misleads future maintainers about the actual scheduler-isolation mechanism. The isolation is now an upstream default, not a Buzz injection. Reverts: [block/buzz#3144](https://github.com/block/buzz/pull/3144) Related: [aaif-goose/goose#10781](https://github.com/aaif-goose/goose/pull/10781) Signed-off-by: Will Pfleger <pfleger.will@gmail.com> |
||
|
|
4a1ebf25c7 |
feat(agent): make Gemini and MLflow-route models usable through databricks_v2 (#3569)
## Summary
Makes Gemini — and every other non-Claude, non-GPT-5 model on the
Databricks MLflow route (`databricks_v2`) — usable in an agent loop.
These are the non-`benchmarks/` changes from
`benchmark/harness-accounting-and-solo`, lifted onto a clean base off
`main` so they can land independently while the harness work continues.
Two defects made these models unusable, one fatal and one silent. Both
live only in `openai_body` / `parse_openai`, which is the
least-exercised of the three `databricks_v2` sub-routes — the
`luna`/`sol` conditions run the Responses route and the `opus`
conditions run the Anthropic route, so **this change is inert for every
model already in use** and only lights up the MLflow path.
## Why a third route at all
`databricks_v2_route_for_model` buckets by model family: `claude*` →
Anthropic Messages, `gpt-5`/code-names → OpenAI Responses, **everything
else → MLflow chat-completions**. Gemini, Qwen, gpt-oss, and friends all
fall through to that third pair — and both bugs below live only there.
## D1 — dropped thought signatures (fatal)
Gemini returns a `thoughtSignature` on every tool call and **requires it
echoed back**. `openai_body` reserialized each call as `{id, type,
function}` only, dropping the field, so the next request 400'd:
```
HTTP 400 Function call is missing a thought_signature in functionCall parts.
```
For a coding agent this fires on the **first** tool call, so the model
never completes a single turn.
**Position is load-bearing.** A four-shape replay probe against the live
gateway established that the signature must sit as a *sibling* of
`function` — nesting it inside `function{}` fails with the *same* 400 as
omitting it. A fix that "preserves the field" without preserving its
position passes a unit test and still 400s.
The fix: `ToolCall` gains `provider_extra: Map<String, Value>`.
`parse_openai` captures every top-level wire key except the three we
model (`id`, `type`, `function`); `openai_body` re-emits them beside
`function`. Keeping *whatever we did not model*, rather than naming
`thoughtSignature`, means the next provider with an opaque per-call
token needs no change here. The Responses and Anthropic replay shapes
are fully modelled, so they pass `Default::default()` and stay
**byte-identical** to before.
### D1b — duplicate tool-call ids (same root cause)
Gemini returns the **function name** as the id, so two parallel calls to
one function arrive sharing an id — and that id is what pairs a
`role:"tool"` result back to its call, making two results
indistinguishable. `dedupe_provider_ids` suffixes collisions
(`get_weather`, `get_weather-2`). Safe because both halves of the
pairing (the assistant `tool_calls[].id` and the result's
`tool_call_id`) are re-emitted from this same value; the provider never
sees its original id again.
## D2 — block-array content discarded (silent, worse than a crash)
`parse_openai` read `content` with `as_str()`, which returns `""` for
anything that isn't a JSON string. Gemini (and Qwen35, gpt-oss) send an
array of typed blocks:
```json
"content": [
{"type": "reasoning", "summary": [{"type": "summary_text", "text": "…"}]},
{"type": "text", "text": "391"}
]
```
So the model answered and the answer was thrown away — no error, no
warning, just a turn that looked like the model had said nothing. On a
benchmark this reads as "Gemini is bad at the task" rather than "buzz
dropped the reply."
`openai_content_parts` now accepts either shape — string as before, or a
block array where `text` blocks concatenate into text and `reasoning`
blocks into reasoning (Gemini nests the prose one level down under
`summary`). Message-level `reasoning_content` / `reasoning` still win
when present, so DeepSeek and vLLM-style hosts are unchanged; block
reasoning is the last fallback.
## Also: a turn-start log line (`buzz-acp` `pool.rs`)
Small, independent observability change that also rides in the
non-benchmark delta: `run_prompt_task` now emits a `pool::prompt` "turn
starting" line, labelled by the same `prompt_label` helper as
`log_stop_reason`, so a log reads as start/stop pairs. An unpaired start
is the only durable evidence that a turn was entered and never returned
— without it, a stalled agent and an agent nobody woke leave identical
(zero-completion) logs.
## Interaction with #3538
#3538 (already merged) rewrote `databricks_v2_route_for_model` to route
by boundary-aware model-family segments. That change and this one touch
**different functions** in `llm.rs` — routing vs. body/parse — and
compose cleanly; the family routing decides *which* pair runs, and this
fixes the MLflow pair it can now select.
## Testing
- `cargo fmt --all -- --check`, `cargo clippy -p buzz-agent -p buzz-acp
--all-targets -- -D warnings` — clean.
- `cargo test -p buzz-agent -p buzz-acp` — all green (304 + 632 lib
tests plus integration suites, 0 failures). Five new tests cover:
block-array text extraction, plain-string regression, passthrough
capture (and non-duplication of the modelled keys), replay position
(`thoughtSignature` beside `function`, not inside it), and id
de-duplication.
- Wire evidence: the four-shape replay table and the reasoning-effort
probe were run against `block-lakehouse-staging` (recorded in the design
doc).
## Relationship to the benchmark branch
The full design write-up (four-shape replay table, position-matters
analysis, effort verification, and open pricing item) lives in
`docs/08-gemini-provider-fixes.md` on
`benchmark/harness-accounting-and-solo`. The benchmark manifests and
endpoint-config entries that exercise these models are separable and
stay on that branch.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Signed-off-by: Atish Patel <atish@squareup.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
||
|
|
9beb3b8c6e |
fix(cli): mask credential env values in --help output (#3570)
clap renders live env var values in help text by default. Three args carrying credentials were exposed this way: - `BUZZ_PRIVATE_KEY` in `buzz-cli` (`crates/buzz-cli/src/lib.rs`) - `BUZZ_AUTH_TAG` in `buzz-cli` - `BUZZ_PRIVATE_KEY` in `buzz-acp` (`crates/buzz-acp/src/config.rs`) Add `hide_env_values = true` to each. Env var names remain visible for discoverability; only their runtime values are withheld from `--help` output. Also adds a regression guard in each crate's test module that walks the clap command tree (recursing into subcommands for `buzz-cli`) and asserts every arg whose env var name contains `KEY`, `SECRET`, `TOKEN`, `PASSWORD`, `CRED`, or `AUTH` has `hide_env_values` set. This prevents future credential-bearing args from being added without the masking in place. Signed-off-by: Will Pfleger <pfleger.will@gmail.com> Co-authored-by: Duncan <dcfd242e557282d7a1e2cf2e6877522682f1e5c6156dc92ca7d90eaedd3b0f95@buzz.block.builderlab.xyz> |
||
|
|
6438dedf83 |
feat(agent): route Claude/GPT model families to their native gateway wire (#3538)
## Summary Databricks v2 chooses the gateway wire format — OpenAI Responses, Anthropic Messages, or MLflow chat — purely from substrings in the endpoint name. There is no family field on the endpoint to key off, so the substring set *is* the routing contract. The matcher only recognised `gpt-5`/`gpt5` and `claude`, which makes correct billing depend on every Claude endpoint happening to be named with the literal string "claude". ## Why this matters Getting a Claude model onto the Anthropic Messages route is exactly what lets buzz attach the `cache_control` breakpoint (the fix in #3463). If a Claude endpoint's catalog name omits "claude" — an alias, a bare `opus-5`, a `goose-opus-5` — it silently falls through to the MLflow (OpenAI-wire) path, where Anthropic prompt caching is **structurally impossible**. The result is the same failure #3463 fixed: 0% cache reads, the full ~10x read discount lost, and no error — a naming convention quietly holding up a billing-correctness invariant. ## What changed `databricks_v2_route_for_model` (`crates/buzz-agent/src/llm.rs`) now matches broader, case-insensitive marker sets: - **Claude → Anthropic Messages:** `claude`, `opus`, `sonnet`, `haiku`, `mythos`, `fable` — the Claude family names and release code names, so a Claude endpoint reaches the cache-capable route regardless of how it's named. - **GPT → OpenAI Responses:** the `gpt` family (now `gpt` on its own, not just `gpt-5`) plus the GPT-5 launch code names `sol`, `luna`, `terra`. OpenAI markers are evaluated first, preserving the prior `gpt-5`-first precedence for any name that could carry both. Names matching neither set still fall through to the MLflow chat route. ## Testing - `cargo fmt`, `cargo clippy -p buzz-agent --all-targets -- -D warnings` — clean. - `cargo test -p buzz-agent` — all green (299 lib + integration suites, 0 failures). The `databricks_v2_routes_by_model_family` test was expanded to cover each new marker, the GPT-5 code names, case-insensitivity, and the unchanged MLflow fallback (including `gemini`). ## Relationship to #3463 #3463 taught the Anthropic path to request caching; this makes sure Claude models actually land on that path. Follow-up still open: surfacing `cache_creation_input_tokens` end-to-end so a persistent `reads == 0 && writes == 0` reveals a disabled cache regardless of which wire a model takes — happy to do that next. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Signed-off-by: Atish Patel <atish@squareup.com> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
|
c405ad1d4b |
feat(agent): fix Anthropic prompt caching with Databricks (+ MCP proxy/TLS passthrough) (#3463)
> On 8 tasks matched by name across the two runs, cost fell $8.36 → $1.77 (4.71×) and wall-clock 12,423 s → 1,085 s (11.45×). ## Summary Two independent, self-contained fixes to `buzz-agent`/`buzz-acp`, split out of the benchmark branch so they can land while the harness work continues: 1. **Request and surface Anthropic prompt caching.** buzz never sent a `cache_control` breakpoint, so on the Databricks Anthropic route `cache_read_input_tokens` was **structurally always 0** and the ~10× cache-read discount was never claimed. This teaches `anthropic_body()` to mark the cacheable prefix, and plumbs the cache split end-to-end so accounting can price it. 2. **Pass proxy + TLS-trust env into MCP tool subprocesses**, so agent tools on a proxy-only host stop reporting a live network as offline. ## Why the caching gap matters The Anthropic Messages API does **not** cache unless the request carries a `cache_control` breakpoint, and the Databricks AI Gateway — a third-party proxy in front of the model, in the same category as Bedrock/Vertex — does **not** auto-cache (only the first-party Anthropic API and Claude-on-AWS do zero-config caching). So every request was billed cold. Measured live against the Databricks gateway (`databricks-claude-opus-5`, 2026-07-28), the same call with and without a single `cache_control` marker: | Run | `input_tokens` | `cache_creation` | `cache_read` | latency | |---|---|---|---|---| | No `cache_control`, two byte-identical calls | 121,625 | 0 | **0** | ~9.3 s | | With one marker — cold (write) | 4 | 121,625 | 0 | 9.3 s | | With one marker — warm (read) | 4 | 0 | **121,625** | **4.5 s** | One marker moved 121,625 tokens from full-price input to a 0.1× cache read and roughly halved latency (a clean, isolated ~2.07× prefill speedup on this single-threaded microbenchmark). The gateway honours `cache_control`; buzz simply never sent it. At fleet scale this was a real budget item. Across matched Terminal-Bench solo sweeps (89 tasks, `-n 20`, before the fix), the two OpenAI-route models independently landed at ~86–87% cache reads — the expected shape for an agentic loop, where system + tools + append-only history repeat every turn — while the Anthropic route returned a hard 0% on every receipt: | Condition | Route | Input tokens | Cache reads | Cost | Cost if uncached | Discount | |---|---|---|---|---|---|---| | luna (`gpt-5-6`) | OpenAI | 20,320,818 | **17.7M (87.0%)** | $6.96 | $22.87 | **3.28×** | | sol (`gpt-5-6`) | OpenAI | 22,312,290 | **19.2M (85.9%)** | $37.07 | $123.35 | **3.33×** | | opus (`claude-opus-5`) | Anthropic | 12,459,822 | **0 (0.0%)** | $81.31 | $81.31 | **1.00×** | Applying luna's measured 87% read rate to the opus token counts at list prices (`input $5/M`, `cached_input $0.5/M`, `output $25/M`) puts the opus run at **~$32.53 vs the $81.31 actually paid — a ~60% overspend on those 49 trials (~$89 on a full sweep)**. That is an upper bound (it prices every cached token at the 0.1× read rate and ignores the 1.25× write premium), and the opus discount is structurally smaller than luna/sol's because opus emits ~3.5× more uncacheable output per trial, which sets a floor on what caching can recover. There is also a plausible **second-order effect**: Databricks appears to meter its per-minute rate limit on *uncached* input tokens, so the missing cache also cost rate-limit headroom — the opus endpoint lost 63% of its trials to fatal 429s while running alone at one-third of a GPT endpoint's raw throughput. This is a hypothesis, not a proven mechanism (the only zero-cache condition is also the only Anthropic endpoint), but it is the reading that explains the throttling with one rule instead of two. ## Post-fix results (provisional — first trials of an in-flight re-run) On 8 tasks matched by name across the two runs, cost fell **$8.36 → $1.77 (4.71×)** and wall-clock **12,423 s → 1,085 s (11.45×)**. | Metric | before (`4a955a858`) | after (`3bef1f6a`) | |---|---|---| | Cache reads as % of input | **0.0%** | **78.7%** (still climbing toward the ~86% steady state) | | `cost_usd_no_cache_discount / cost_usd` | **1.00×** | **2.18×** (tracking the projected ~2.5×) | | Trials with a fatal 429 (same `-n 20`) | **63%** | **15–19%** | To be clear about attribution: **~2× of that is the clean prefill saving from caching itself**; the rest is second-order — cached requests burn far less rate-limit budget, so they stall less and redo less destroyed work. The 11.45× is a system-level result specific to this throttled workspace, not a caching benchmark. Quality held (7/8 solved in each run). A controlled low-`-n` A/B (neither arm hitting a 429), which the `BUZZ_AGENT_PROMPT_CACHING` opt-out exists to enable, is still owed before this becomes a published claim. ## What changed ### 1. Request caching (`llm.rs`, `config.rs`) `anthropic_body()` emits ephemeral `cache_control` breakpoints, gated by `BUZZ_AGENT_PROMPT_CACHING` (**default on**, `=0` to opt out): - **Static prefix** — marker on the `system` block. Prefix order is `tools → system → messages`, so this single marker caches **tools + system** together. Byte-identical on every turn of a run, and survives a context handoff (system/tools come from cfg/mcp, not `self.history`). - **Rolling tail + leapfrog** — marker on the last block of the last **two** messages. The append-only history re-reads the prior turn's prefix from cache; marking two messages (not one) keeps consecutive breakpoints inside Anthropic's **20-block lookback window** even as tool parallelism rises, avoiding a silent full-price miss. An empty system prompt stays a bare string (Anthropic rejects empty text blocks), and below-threshold prefixes are silently not cached, so the flag is safe on by default. ### 2. Surface the cache split end-to-end — the plumbing (`types.rs`, `llm.rs`, `agent.rs`, `lib.rs`, `usage.rs`, `acp.rs`) This is the part that makes gaps like the one above **visible** instead of silent. A consumer that prices all of `input_tokens` at the full rate can't tell a route that's caching from one that isn't — the total looks right either way. So: - `LlmResponse` gains `cached_input_tokens` (a **subset** of `input_tokens`, never an addition); `parse_anthropic` / `parse_openai` / `parse_responses` each populate it. - A `usage_first()` helper reads the cache count wherever a provider hides it — flat `cache_read_input_tokens` (Anthropic), `prompt_tokens_details.cached_tokens` (OpenAI chat), `input_tokens_details.cached_tokens` (Responses) — taking the **first present value, never a sum**. Reading only flat keys is exactly why the OpenAI route's nested `cached_tokens` had *also* been going unclaimed: `prompt_tokens` is already inclusive, so the total looked correct while the discount silently went unreported. - The per-turn/per-session accumulators and the goose `usage_update` payload now carry `accumulatedCachedInputTokens`; `buzz-acp` deserializes it (`serde` default `0` for goose, which doesn't send it) and logs `cached=<n>`. ### 3. Fix a Databricks MLflow-route double-count (`llm.rs`) The Databricks MLflow route reports the flat Anthropic-spelled `cache_read_input_tokens` *alongside* an already-inclusive `prompt_tokens`, so the old code summed them and nearly doubled the count — inflating both the context-budget gate and cost. `openai_chat_input_tokens()` now reads `prompt_tokens` alone. Verified on a live `databricks-glm-5-2` response where `prompt_tokens + completion == total` proves inclusivity. (Anthropic's native route genuinely *excludes* the cache fields and is still summed — the two never collide, because `claude*` models route to the Anthropic path.) ### 4. Proxy + TLS-trust passthrough into MCP tools (`mcp.rs`) — independent fix `buzz-agent` `env_clear()`s each MCP child, and the allowlist carried no proxy/TLS vars. On a proxy-only host that doesn't degrade the tools, it **blinds** them: apt, curl, pip, git connect directly, the egress firewall resets the socket, and the agent reports "Connection reset by peer" — indistinguishable from a genuinely offline task. Adds both spellings of `HTTP(S)_PROXY`/`NO_PROXY`/`ALL_PROXY` (curl/git read lowercase; Go/Python read uppercase; libcurl ignores uppercase `HTTP_PROXY`) plus `SSL_CERT_FILE`/`SSL_CERT_DIR` for TLS-terminating proxies that present their own CA. ## Testing - `cargo fmt --all -- --check`, `cargo clippy -p buzz-agent -p buzz-acp --all-targets -- -D warnings` — clean. - `cargo test -p buzz-agent -p buzz-acp` — **all green** (632 + 299 lib tests plus integration suites, 0 failures). New tests cover: the three breakpoints and the disabled/empty-system/single-message edge cases; nested-vs-flat cache parsing for all three routes; the Databricks inclusive-`prompt_tokens` fix; wire deserialization of `accumulatedCachedInputTokens`; and the proxy/TLS passthrough allowlist. - Pre-push lefthook suite green (branch-skew, rust-tests, test, desktop-check/test/tauri). ## Relationship to the benchmark branch These are the non-`benchmarks/` changes from `benchmark/harness-accounting-and-solo`, lifted onto a clean base off `main` so they can merge independently. 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Signed-off-by: Atish Patel <atish@squareup.com> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
|
6300a6b1d0 |
fix(acp): per-runtime env defaults at spawn — isolate Hermes from configured MCP startup (#3420)
## Summary - add a generic per-runtime env-defaults table, `config::default_agent_env()`, mirroring the existing `default_agent_args()` / `codex_network_env()` precedent, and merge it once in `AcpClient::spawn` with the established precedence: **runtime defaults < persona `extra_env` < inherited parent env** - first (and only) row: Buzz-owned Hermes processes get `HERMES_ACP_SKIP_CONFIGURED_MCP=1`, so Hermes does not preload unrelated profile-configured MCP servers before answering ACP `initialize` (fixes the 10s model-discovery timeout in #3355 — Buzz supplies session MCP servers explicitly through `session/new`, per Hermes's documented host-integration contract for this variable) - normalize Windows `.cmd`/`.bat` shims alongside `.exe` in `normalize_agent_command_identity` (npm installs resolve to those wrappers) - switch the `extra_env` parent-presence check from `var()` to `var_os()` so non-UTF-8 parent values are honored Replaces the runtime-specific approach in #3386: same behavior, but the mechanism is generic runtime spawn metadata in `config.rs` rather than a Hermes/ACP special case in `acp.rs`, and the seam covers every launch path (Desktop spawn, `buzz-acp models`, CLI) because they all funnel through `AcpClient::spawn`. ~15 lines of production code. Fixes #3355 ## Testing - `cargo test -p buzz-acp` — **639 passed, 0 failed** (full package, includes the new `default_agent_env_recognizes_hermes_identities` unit test and `spawn_applies_runtime_env_defaults_with_extra_env_precedence` integration test covering default injection, extra_env override, and non-Hermes exclusion) - `cargo fmt --all -- --check`, `cargo clippy -p buzz-acp --all-targets -- -D warnings` — clean - live-local with real Hermes v0.19.0 (`hermes-acp`): `buzz-acp models` returned **13 models / currentModelId in 2.6–3.0s** (was a 10.0s timeout on the first cold run without isolation); a wrapper probe confirmed the child received `HERMES_ACP_SKIP_CONFIGURED_MCP=1` by default and `0` when the parent env set it explicitly (operator wins) - lefthook pre-push suite green: rust-tests, desktop-check, desktop-test, desktop-tauri-test, mobile-test, branch-skew No UI changes; subprocess environment behavior only. Signed-off-by: npub1qyvc0c5kl4gqv2fd97fsk46tu378sqgy35vc83rvgfwne90sel7s0ed67d <011987e296fd5006292d2f930b574be47c7801048d1983c46c425d3c95f0cffd@buzz.block.builderlab.xyz> Signed-off-by: Tyler Longwell <tlongwell@block.xyz> Co-authored-by: npub1qyvc0c5kl4gqv2fd97fsk46tu378sqgy35vc83rvgfwne90sel7s0ed67d <011987e296fd5006292d2f930b574be47c7801048d1983c46c425d3c95f0cffd@buzz.block.builderlab.xyz> Co-authored-by: Tyler Longwell <tlongwell@block.xyz> Co-authored-by: mr-r0b0t.eth <adam.manning@pro-serveinc.com> |
||
|
|
22be8bb351 |
fix(relay): avoid subscription lock inversion (#3413)
## Summary - drop the `subs` DashMap guard before mutating subscription indexes - snapshot fan-out candidate vectors so index guards are dropped before looking up `subs` - add concurrent fan-out/replacement regression coverage ## Why `fan_out_scoped` previously held an index guard while `push_match` acquired `subs`, while CLOSE and same-ID replacement held `subs` while removing from an index. The reverse ordering made an AB/BA deadlock reachable and could synchronously park all Tokio workers. ## Validation - `rustup run 1.95.0 cargo test -p buzz-relay` — 769 library tests passed, 33 ignored; 11 binary tests passed; doc tests passed - push hooks with pinned Rust 1.95 — branch-skew, repository Rust suites, and desktop Tauri suite passed - `git diff --check` ## Residual risk Fan-out now clones bounded candidate vectors before matching. This adds allocation/copy cost proportional to the indexed candidate set, in exchange for eliminating nested DashMap guards. This fixes the concrete lock cycle but does not prove every observed production wedge had this cause. --------- Signed-off-by: npub12gtutshhh76rx0jx697f32f9tffd4hhp3hx58fp4x6u4uemkm7sqf8f757 <5217c5c2f7bfb4333e46d17c98a9255a52dadee18dcd43a43536b95e6776dfa0@buzz.block.builderlab.xyz> Signed-off-by: npub1jh9wn95s0472h86ahapupaf7m6kx4v9sx2n0atj2hltcfer8k06s5n3pyf <95cae996907d7cab9f5dbf43c0f53edeac6ab0b032a6feae4abfd784e467b3f5@buzz.block.builderlab.xyz> Co-authored-by: npub12gtutshhh76rx0jx697f32f9tffd4hhp3hx58fp4x6u4uemkm7sqf8f757 <5217c5c2f7bfb4333e46d17c98a9255a52dadee18dcd43a43536b95e6776dfa0@buzz.block.builderlab.xyz> Co-authored-by: npub1jh9wn95s0472h86ahapupaf7m6kx4v9sx2n0atj2hltcfer8k06s5n3pyf <95cae996907d7cab9f5dbf43c0f53edeac6ab0b032a6feae4abfd784e467b3f5@buzz.block.builderlab.xyz> |
||
|
|
90e058ebf6 |
feat: add explicit entry for claude-opus-5 in model config (#2831)
Fixes #2787 - Added `claude-opus-5` to `config.rs` model classification and adaptive effort helpers. - Updated fixture test configurations to cover `claude-opus-5`. - Verified with `cargo test` and JS unit tests. Signed-off-by: Apurva Shaw <apurvashaw@Apurvas-MacBook-Air.local> Co-authored-by: Apurva Shaw <apurvashaw@Apurvas-MacBook-Air.local> |
||
|
|
f25e6dd6aa |
feat(acp): steer claude-code and codex agents via _session/steering (#3007)
Mid-turn steering was reachable only through goose's
`_goose/unstable/session/steer`, which requires an `expectedRunId`
sourced from `_meta.goose.activeRunId`. claude-agent-acp and codex-acp
never emit a run id, so every mid-turn mention to those harnesses bailed
at the run-id guard before writing a byte and degraded to cancel +
merge, destroying in-flight tool calls.
Both adapters ship `_session/steering` (params `{sessionId, prompt}`,
result `{outcome}`) and advertise it as `_meta.steering.supported` on
the `initialize` response. This adds it as a second steer transport
selected at write time, reusing the existing withhold/release, ack
routing, and cancel+merge fallback machinery unchanged.
## Transport selection
| `active_run_id` | `steering_supported` | Transport |
|---|---|---|
| `Some(run_id)` | any | `_goose/unstable/session/steer` +
`expectedRunId` (unchanged) |
| `None` | `true` | `_session/steering` with `{sessionId, prompt}` |
| `None` | `false` | ack `ExpectedRunIdMissing`, write nothing
(unchanged) |
goose keeps priority when both are present — `expectedRunId` is strictly
more precise about *which* run is being steered.
## Two load-bearing safety properties
**The advertised capability is the only gate — never error-code
probing.** codex-acp's `extMethod` answers unrecognized extension
methods with a bare `{}`, which is a JSON-RPC *success* rather than
`-32601`. Buzz maps a steer success to `queue.remove_event`, so probing
an unknown method would silently delete the user's message with no
error, no fallback, and no log line.
**An `outcome` must be positively recognized.** Only `injected` and
`startedNewTurn` count as delivery. Anything else — codex's `failed`, an
unknown value, or a missing `outcome` entirely — is
`SteerError::OutcomeRejected`, which releases the withheld event and
fires the cancel+merge fallback. This makes the silent-loss path above
unreachable even if an adapter mis-advertises.
`startedNewTurn` acks `Success`, because the message really was
delivered and must not be redelivered, but deliberately does **not**
renew the read loop's hard deadline: the turn Buzz was awaiting had
already settled, and renewing would extend the clock on a finished turn.
## Notes for reviewers
- `SteerError::OutcomeRejected` needs no new arm in the
`PoolEvent::SteerAck` match — the existing catch-all
`Ok(SteerAck::Err(_)) => (true, false, true)` already gives release +
fallback, and the two `AgentError` arms above it match that variant
specifically, so they do not shadow it.
- Comments that described the old goose-only "try-and-tolerate" `-32601`
behavior are corrected; that assumption was never valid for codex-acp.
- No CI job runs `buzz-acp` tests. The full package suite was run
locally: **617 passing, 0 failing**.
Signed-off-by: Will Pfleger <pfleger.will@gmail.com>
Co-authored-by: npub1mn7jgtj4w2pd0g0zeuhxsa6jy6p0rewxz4kujt98my82ahfmp72sxjexk7 <dcfd242e557282d7a1e2cf2e6877522682f1e5c6156dc92ca7d90eaedd3b0f95@buzz.block.builderlab.xyz>
|
||
|
Will PflegerandGitHub
|
1d4f97b959 |
fix(acp): disable goose cron scheduler in managed agent children (#3144)
A Buzz install with a scheduled goose recipe fires each cron entry once per `goose acp` child instead of once, because every child unconditionally starts its own cron scheduler over the shared `~/.local/share/goose/schedule.json`. With a pool of N children per harness and multiple harnesses, one scheduled recipe fans out to N × harness_count executions — each running under the managed agent's identity rather than the operator's, and racing the operator's own standalone goose over the same schedule file. This injects `GOOSE_ACP_SCHEDULER_DISABLED=true` into every child spawned by `AcpClient::spawn`, so a managed agent never owns the operator's cron schedule. ## Placement The `cmd.env` call is set last — after the `extra_env` operator-wins loop and after the `CODEX_CONFIG` merge — deliberately with no escape hatch. Managed children not running the operator's schedule is a correctness invariant rather than an operator-tunable default, so the injection must beat both a conflicting persona `extra_env` entry and any value inherited from the parent process. It is injected for all agents, not just goose. Agent builds that don't recognize the variable ignore it. ## Sequencing The goose-side flag that reads this variable and skips scheduler startup lands separately (repo TBD). Until it does, this change is a forward-compatible no-op: it sets an environment variable nothing currently reads. Merging it first means no coordinated release is needed — the fix takes effect as soon as the goose side ships. Related: https://github.com/aaif-goose/goose/pull/10738 Signed-off-by: Will Pfleger <pfleger.will@gmail.com> |
||
|
|
60158fce3e |
feat(cli): add users set-status command for NIP-38 profile status (#3253)
## Summary The desktop client renders a persistent user status (NIP-38 kind:30315, `d:general`) as the status line on profiles, but the CLI had no way to set it — only ephemeral presence (`set-presence`, kind:20001). Integrations that want a scriptable, durable status line (for example a now-playing music bridge that shows the current TIDAL track on a profile) had no entry point. ## Screenshots <img width="1455" height="960" alt="1" src="https://github.com/user-attachments/assets/f1669ec6-212b-4f6e-ad53-07df9aacffc9" /> <img width="1455" height="960" alt="2" src="https://github.com/user-attachments/assets/5bf70f47-e5b5-4eb0-a426-b5f1ef90d2ec" /> This adds: ```bash buzz users set-status --text "Working on the relay" --emoji "🔧" buzz users set-status --text "" --emoji "🎶" # intentional emoji-only status buzz users set-status --clear # removes the status ``` - Signs and submits the replaceable kind:30315 event via the HTTP bridge (no WS needed — unlike presence, user status is a stored event). - Uses the `d:general` coordinate the desktop client already reads for the profile status line, and the same `emoji` tag shape `SetStatusDialog` publishes. - Event construction lives in `buzz_sdk::build_user_status()`, keyed off `buzz_core::kind::KIND_USER_STATUS`, so the CLI command is a thin sign/submit wrapper. Text and emoji are trimmed; a blank emoji is omitted rather than emitted as an empty tag. - Clearing is the explicit `--clear` flag, mutually exclusive with `--text`/`--emoji`. It publishes an empty-content event carrying only `d:general`, which the desktop treats as no status. `--text ""` with an `--emoji` is an emoji-only status, not a clear. --------- Signed-off-by: Kagan Yaldizkaya <kagan@squareup.com> Signed-off-by: Will Pfleger <pfleger.will@gmail.com> Co-authored-by: Will Pfleger <pfleger.will@gmail.com> |
