mirror of
https://github.com/nexu-io/open-design.git
synced 2026-10-27 09:51:12 +01:00
common-sphere
186
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 pftom
|
31c0bd159f |
fix(research): require canonical SHA on install, validate memory tags
- installHubEntry now requires entry.source.commitSha to match FULL_COMMIT_SHA_PATTERN (40 hex chars). The shorter prefix shape was a trust gap: an in-process caller could craft an entry whose commitSha matched the approval string by luck, pass the equality check, and trigger a re-fetch with ?ref=<short> that GitHub resolves through branch/tag names. - createMemory now validates input.tags as a bounded array of slug strings (max 16 entries, 64 bytes each, 512 bytes total) and rejects control characters and prompt-frame tokens, so a future memory route cannot bypass the 8KB content cap by smuggling instruction-like strings through tags that buildMemoryPrefix renders into the prompt JSON. - Added regression tests for both gates. Generated-By: looper 0.6.1 (runner=fixer, agent=claude-code) |
||
|
pftom
|
8ff084cf44 |
fix(research): thread server-owned token to installHubEntry re-fetch
`installHubEntry` re-fetches the manifest at the approved commit SHA before writing, but the call into `fetchManifestRawAtSha()` did not pass any token. `HubEntry.source` deliberately records `repo/path/commitSha` without a token (forwarding an entry must not exfiltrate credentials), so for a hub listed through `HubConfig.token` / `OD_HUB_TOKEN` the re-fetch hit the GitHub API anonymously and returned 404 — breaking the in-process install path documented as usable for entries from `listHubEntries()` against a private hub (PR #617 review, P1). `InstallOptions` now carries an optional server-owned `token`. The install path prefers `opts.token` and falls back to `OD_HUB_TOKEN` when the caller leaves it unset, mirroring the env-var contract `listHubEntries()` already honors. The token is intentionally kept out-of-band on the options object so a forwarded `HubEntry` still cannot carry credentials across a process boundary; callers that don't already hold a server-owned token should route through `installHubManifest`, which takes the token on the selector instead. Adds two regression cases to `apps/daemon/tests/research-skill-hub-install.test.ts`: - `threads opts.token through the re-fetch so private-hub entries install` — the mock returns 404 unless the bearer token matches, so the test fails if the install path forgets to thread the token. - `falls back to OD_HUB_TOKEN when opts.token is not supplied` — covers processes configured the same way as the listing call site. Generated-By: looper 0.4.0 (runner=fixer, agent=claude-code) |
||
|
pftom
|
624820f192 |
fix(research): partition skill-hub cache, re-fetch on install, bound audit lines
Addresses four PR #617 review threads: - listHubEntries cache key now partitions on a per-process HMAC fingerprint of the bearer token (or the literal "public" sentinel for tokenless requests). This stops a tokened request that surfaces private hub manifests from being served back to a later tokenless request with the same (repo, ref, namespace), where `entry.raw` is prompt material. - installHubEntry no longer trusts caller-supplied `entry.raw`. After the SHA equality check it re-fetches the manifest from `/repos/.../<entry.source.path>?ref=<entry.source.commitSha>` and writes only the freshly fetched bytes; if those bytes drift from `entry.raw` the install is rejected. A new `installHubManifest` entry point takes a server-owned selector `{ namespace, name, repo, pinnedSha }` and is the recommended surface for any future `POST /api/skill-hub/install` route — HTTP bodies never carry manifest bytes through the writer. - The audit-log redaction pass now caps array length, object key count, and the final serialized JSONL line. Items past the per-container caps are replaced with a single truncation marker; lines past the byte cap collapse `detail` to a `__truncated` marker carrying the original size, so a pathological `detail` cannot block the daemon on synchronous JSON.stringify / appendFileSync. - The Skill Hub example in `docs/research/openwork-aionui-alaude- integration.md` now mirrors the safe install path the starter enforces: pinned-SHA listing, `installHubManifest` taking a selector + approval, re-fetch inside the install path, the quarantine/activation boundary, and SHA-pinned sibling-asset fetches. Adds `apps/daemon/tests/research-audit-log-bounds.test.ts` covering the new caps, and extends the install test with a re-fetch / drift case plus selector-based install coverage. Generated-By: looper 0.4.0 (runner=fixer, agent=claude-code) |
||
|
pftom
|
31399fefed |
fix(research): resolve hub pinnedSha to canonical commit SHA via GitHub API
`COMMIT_SHA_PATTERN` was treating any 7–40 hex string as immutable provenance, but GitHub's `?ref=` parameter also accepts branch and tag names. A hub repo could create a mutable branch/tag named like a short SHA, pass the regex, fetch mutable-ref bytes, and then satisfy `installHubEntry` because `entry.source.commitSha` and `approval.pinnedSha` would both be the caller-supplied string (PR #617 review, P1). `listHubEntries` now resolves the caller-supplied SHA prefix through `GET /repos/{owner}/{repo}/commits/{input}` before any listing happens. The canonical 40-char `.sha` is required to share the input's lowercase hex prefix; otherwise GitHub resolved the ref through a branch/tag and we refuse to treat it as immutable provenance. The canonical full SHA is then used as the `?ref=` value and stamped on every `HubEntry.source.commitSha`, so `installHubEntry`'s exact-match equality compares 40-char canonical values on both sides. Adds `apps/daemon/tests/research-skill-hub-resolve.test.ts` with three cases: hex-shaped pinnedSha that resolves through a branch is rejected before any Contents API call; a real prefix that resolves to a sharing commit is accepted; and the listing/Contents API both use the canonical 40-char SHA, never the caller's raw 7-char input. Generated-By: looper 0.4.0 (runner=fixer, agent=claude-code) |
||
|
pftom
|
8f66221e0b |
fix(research): tie skill-hub install bytes to the approved commit SHA
`installHubEntry` previously accepted any SHA-shaped `approval.pinnedSha` without verifying that the bytes in `entry.raw` came from that commit. Because `listHubEntries` fetched through the configured mutable branch ref and `HubEntry.source` carried no immutable provenance, a caller could list from `main`, pass any SHA-shaped string with `userApproved: true`, and still install unpinned branch content into active prompt directories. This change carries provenance end-to-end: - `HubConfig` accepts `pinnedSha` (or `OD_HUB_PINNED_SHA`); when set, `listHubEntries` / `readDirEntry` / `readFileEntry` route the GitHub Contents API through `?ref=<sha>` and stamp every returned `HubEntry.source.commitSha` with that SHA. - The list cache key now hashes the resolved ref (SHA or branch) so a pinned listing can never collide with a branch-tip listing. - `installHubEntry` now refuses any entry whose `source.commitSha` is missing (mutable-branch listing) or does not match `approval.pinnedSha` (case-insensitive exact equality, no prefix matching). The planning comment block is updated to reflect that the SHA-anchored install path is now enforced rather than future-tense. Adds `apps/daemon/tests/research-skill-hub-install.test.ts` covering: mutable-branch entry rejected, mismatched approval SHA rejected, matched SHA accepted, case-insensitive equality, short-prefix SHA rejected, untrusted `userApproved: false`, and malformed approval SHA. Generated-By: looper 0.4.0 (runner=fixer, agent=claude-code) |
||
|
pftom
|
02cd4c61e7 |
fix(research): expand skill-hub mode list and parse block-scalar descriptions
Frontmatter validation in the research skill-hub starter now mirrors the shapes already used by `skills/*/SKILL.md`: - `ALLOWED_OD_MODES` adds `image`, `video`, `audio`, and `utility` so media and utility skills (e.g. `image-poster`, `video-shortform`, `audio-jingle`, `pptx-html-fidelity-audit`) are no longer rejected by a hub mirroring the catalog. - The minimal YAML parser now reads `description: |`, `description: |-`, `description: >`, and `description: >-` block scalars instead of storing the literal `|` / `>`. Continuation lines are joined with `\n` and right-trimmed, matching `apps/daemon/src/frontmatter.ts`. - Adds `apps/daemon/tests/research-skill-hub-frontmatter.test.ts` with fixture coverage against representative live skills for every currently-used `od.mode` value plus inline cases for `>` / `|-`. Generated-By: looper 0.4.0 (runner=fixer, agent=claude-code) |
||
|
pftom
|
c706866068 |
fix(research): redact camelCase secrets, gate hub install, daemon-side browser MCP
- audit-log: redact camelCase secret keys (`accessToken`, `refreshToken`, `clientSecret`, `sessionId`, etc.) by inserting separators at case transitions before pattern test (PR #617 review, P2). - skill-hub: require explicit `InstallApproval { pinnedSha, userApproved }` on `installHubEntry` and document the trust boundary — manifest validation alone is not a trust step (PR #617 review, P1). - research doc: rewrite browser-use plan to remove the false "all agents are sandboxed" / "MCP tool consent already standard" premises; require daemon-owned MCP proxy with default-deny, transport-level read-only allowlist, and per-tool UI approval (PR #617 review, P1). Generated-By: looper 0.4.0 (runner=fixer, agent=claude-code) |
||
|
pftom
|
1c37080ed9 |
fix(research): bound audit redaction, reader limit, and memory prefix
Generated-By: looper 0.4.0 (runner=fixer, agent=claude-code) |
||
|
pftom
|
0a12cdb1c9 |
fix(research): accept existing DESIGN.md / craft shape in skill-hub
Address PR #617 review comment PRRC_kwDOSOgY8s6-ULyy: the previous frontmatter validator rejected the 140+ design systems and the craft modules Open Design already ships, since those files use a plain Markdown shape (`# <Heading>` plus `> Category:` blockquote) rather than YAML frontmatter. A hub repo mirroring the existing artifacts would have returned no entries from `listHubEntries()` and thrown from `installHubEntry()`, which defeats the starter's stated purpose. Add a heading-based fallback for `design-systems` and `craft`: - `deriveHeadingDescription(raw)` reads the first `# ` heading and the first blockquote block (cap 280 chars), stripping any frontmatter that may have been present. - `resolveManifest(namespace, raw, manifestId)` keeps `skills` strict (frontmatter required, since all 65+ shipped skills carry `name` / `description`) but accepts either frontmatter or the plain-Markdown shape for `design-systems` and `craft`. The manifest id is still re-validated against the namespace's slug rules to keep path-traversal guards intact, and any `name` or `od.mode` present in frontmatter must still be valid. - `readDirEntry`, `readFileEntry`, and `installHubEntry` route through `resolveManifest` instead of `validateFrontmatter`. `validateFrontmatter` itself stays as-is (still strict, still exported) so the v0 schema remains documented for hubs that opt into frontmatter. Sanity-checked against `design-systems/apple/DESIGN.md`, `design-systems/atelier-zero/DESIGN.md`, and `craft/accessibility-baseline.md`: all three now resolve to non-null entries with sensible derived descriptions, while `resolveManifest('skills', appleRaw, 'apple')` still returns null. Generated-By: looper 0.4.0 (runner=fixer, agent=claude-code) |
||
|
pftom
|
897e477277 |
fix(research): reject "."/".." project ids and split audit-file resolver
Address two follow-up review comments on PR #617: - isSafeProjectId: explicitly reject "." and "..". The existing character allowlist `[A-Za-z0-9._-]{1,128}` accepted bare-dot ids, and the defense-in-depth check still permitted `dir === projectsRoot`, so `projectId="."` would have written `audit.jsonl` directly under `<dataDir>/projects/`. Tighten the strict-prefix check to require a proper subpath of the projects root. - Split path resolution from directory creation per reviewer's suggested shape: rename `resolveAuditPath` to `auditFilePath` (pure resolver, returns the full file path as a `string`) and have `ensureAuditFile` call it then `mkdirSync(path.dirname(file))`. `appendAuditEntry` continues to call ensure, `readAuditEntries` calls the pure resolver only — reads cannot create directories. Generated-By: looper 0.4.0 (runner=fixer, agent=claude-code) |
||
|
pftom
|
8d4e940183 |
fix(research): harden memory/audit/hub starters per PR review
Address P1/P2 security review feedback on the unwired research starters under apps/daemon/src/research/: - memory-schema.ts (P1 prompt injection): reject control characters and prompt-frame tokens at create/update time, render entries as <-escaped JSON in `buildMemoryPrefix`, and reword the wrapper to mark memory as user-provided context rather than authoritative instructions. Also clamp the LIMIT in `listMemoriesForContext` to a finite integer and bind it as `LIMIT ?` instead of interpolating. - audit-log.ts (P1 path traversal + P2 redaction no-op): split path resolution from `mkdirSync` so reads are side-effect-free; add a recursive redactor that strips token / api_key / authorization / cookie / password / private_key / session_id / bearer / x-api-key fields before append and on export, with a per-string length cap and bounded recursion depth. - skill-hub.ts (P2 download cap + frontmatter validation): replace `arrayBuffer()` with a streamed reader that bails on Content-Length over the cap and cancels mid-download once accumulated bytes exceed it; replace the description-only regex with a minimal frontmatter parser plus per-namespace validation (name slug + manifest-id match, description required, `od.mode` allowlist) applied on both list and install paths. No runtime wiring changes; module remains unimported by server.ts. Generated-By: looper 0.4.0 (runner=fixer, agent=claude-code) |
||
|
pftom
|
f7ffce26eb |
fix(research): guard project id in audit-log starter
Validate `projectId` against the same `^[A-Za-z0-9._-]{1,128}$` allowlist
that `apps/daemon/src/projects.ts#isSafeId` uses, and verify the resolved
audit directory stays under `<root>/projects/`, before `mkdirSync()`.
Prevents a path-traversal bug if this starter is wired to the planned
`GET /api/projects/:id/audit` route with a malformed `:id`.
Generated-By: looper 0.4.0 (runner=fixer, agent=claude-code)
|
||
|
pftom
|
4f23f34bf2 |
fix(research): respect includeArchived; install craft at root
Generated-By: looper 0.4.0 (runner=fixer, agent=claude-code) |
||
|
pftom
|
2f55171595 |
docs(research): openwork / AionUi / alaude-desktop integration analysis
Adds a deep capability survey of three external Agent-client projects and a minimum integration starter for Open Design. - docs/research/openwork-aionui-alaude-integration.md: full analysis, capability matrix, P0/P1/P2 extraction list with rationale, roadmap, and pitfalls. Each item carries "why useful" + "why open-design needs it" + minimum landing form. - apps/daemon/src/research/: standalone TypeScript starters for the three highest-ROI items (memory schema + injection helper, audit log JSONL writer/reader, GitHub-as-registry skill hub fetcher). NOT wired into server.ts; subdir README states the integration path for follow- up PRs (one PR per item, with contracts DTOs and tests). Validation: pnpm guard + pnpm --filter @open-design/daemon typecheck. |
||
|
|
2036ce0a8e |
feat(web): add Cmd/Ctrl+P quick file switcher (#556)
* feat(web): add Cmd/Ctrl+P quick file switcher A keyboard-driven file palette overlaid on the workspace. Press Cmd/Ctrl+P anywhere in the project view; type to fuzzy-filter the file list, ↑/↓ to navigate, Enter to open in a tab, Esc to dismiss. With an empty query the palette surfaces recents (per-project, localStorage) followed by the rest of the file list sorted by mtime. Adds: - apps/web/src/components/QuickSwitcher.tsx: palette UI and matcher - apps/web/src/quickSwitcherRecents.ts: per-project recents store - index.css: scoped .qs-* styles using existing design tokens - i18n: 6 new keys translated across all 16 locale files Wires into FileWorkspace's existing openFile() so recents and tab state behave identically to opening from DesignFilesPanel. Capture-phase keydown beats the browser's print dialog. No backend changes; uses the files prop already passed to FileWorkspace. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(web): address QuickSwitcher review feedback Three fixes from the PR review: - z-index: bump .qs-overlay from 200 to 1500 so the palette renders in the modal tier (alongside prompt-template-modal-overlay) instead of behind context menus and popovers (which sit at 200). - Arrow-key guard: skip setCursor when matches is empty. Without this, pressing ↓ on a no-results query set the cursor to -1, making the highlight selector miss every row on the next render. - Tests: add 19 unit tests covering scoreMatch ranking tiers, render output (empty state / row count / kbd hints / placeholder), and the full recents lifecycle (cap at 6, dedupe-on-push, corrupt-JSON recovery, per-project scoping, quota-exceeded no-op). Vitest stays on the node env via a small in-memory localStorage stub. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(web): QuickSwitcher review — wrap, IME, platform gate Three follow-ups from @mrcfps's review on #556: - ArrowUp/ArrowDown now wrap at list bounds (last → first, first → last) via modulo arithmetic in a new pure helper `nextCursor(current, total, direction)`. Previously they clamped, which contradicted the wrap behavior the PR test plan promised. Pulled into a pure function so boundary cases are unit-testable without simulating keyboard events. - Palette's onKeyDown now early-returns on `e.nativeEvent.isComposing`, so users typing CJK file names through an IME keep ↑/↓/Enter for candidate navigation instead of having them steered by the palette. The global Cmd/Ctrl+P opener already had the equivalent guard. - Global keydown is now platform-gated: macOS responds only to metaKey, win/linux only to ctrlKey. Previously both fired everywhere, which meant Ctrl+P on macOS was stealing native readline "previous line" in text fields (and the chat composer). Tests: +6 unit tests for `nextCursor` covering forward/backward wrap, mid-list moves, empty list (no division-by-zero), and single-item no-op. Suite now 258 passing (up from 252). Verified live: ↓ from last row → first row; ↑ from first row → last row, in a mocked-project Playwright harness. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com> |
||
|
ferasbusiness666andGitHub
|
e8b63ecec1 | feat: add resizable chat panel (#563) | ||
|
bojie.hbjandGitHub
|
29e1ed5f28 | fix: 修复设计文件列表行菜单被裁剪的问题 (#561) | ||
|
TomandGitHub
|
6b2792b03a |
fix(daemon): remove --no-session from pi adapter to persist session files (#557)
* fix(daemon): remove --no-session from pi adapter to persist session files The pi agent was the only adapter explicitly passing `--no-session` in its `buildArgs`, preventing pi from writing session files. All other adapters either run in single-shot mode by design or use the ACP JSON-RPC session lifecycle without suppressing persistence. Removing `--no-session` lets `--mode rpc` retain its default behavior of writing session state, which is needed for multi-prompt continuity and matches the rest of the harness ecosystem. * test(daemon): add pi buildArgs regression tests; fix docs for --no-session removal - Adds test for pi.buildArgs base shape: returns ["--mode", "rpc"] and does not include --no-session (prevents regression). - Adds test for --model and --thinking option passthrough. - Updates pi-rpc.ts lifecycle comment to remove [--no-session]. - Updates README.md and all localized READMEs to reflect the corrected pi CLI invocation. |
||
|
Marc ChanandGitHub
|
0953ab58a2 |
fix(web): todo card dark mode styling (#526)
* fix(web): improve todo card dark mode * fix(web): address todo card contrast feedback Generated-By: looper 0.5.5 (runner=fixer, agent=opencode) |
||
|
lefarcenandGitHub
|
963bbf2500 | release: Open Design 0.4.0 (#454) | ||
|
ChildhoodAndyandGitHub
|
009d7a5478 |
refactor(daemon): eliminate duplicate dist tree from two-tsconfig build (#553)
Move sidecar source under src/ so a single tsconfig produces all daemon
output. Removes the parallel dist/src/ tree that was emitted by
tsconfig.sidecar.json (it included src/**/*.ts to type-check the
`../src/server.js` cross-tree import).
Build now emits:
- dist/<flat> (cli.js, server.js, app-version.js, ...)
- dist/sidecar/{index,server}.js
`dist/sidecar/server.js` reaches the main daemon via `../server.js`
instead of `../src/server.js`, so there is no second copy of the source
tree in the published tarball.
Background — issue #534 (already fixed by #537):
The packaged Settings → About panel showed 0.0.0 because the sidecar
chain loaded the duplicated `dist/src/app-version.js`, where the fixed
`new URL('../package.json', import.meta.url)` resolved to a non-existent
`dist/package.json`. #537 patched the symptom by walking parents until a
real `package.json` is found and by writing `appVersion` into the Linux
packaged config. Both stay in place — they're sound defenses — but the
underlying duplicate-emit was never addressed; any future relative
resource lookup (templates, schemas, prompts) anchored on
`import.meta.url` would have hit the same trap.
This change removes the trap.
|
||
|
soulmeandGitHub
|
6b7a40e5c3 | Fix file tab wheel scrolling (#549) | ||
|
Justin GaoandGitHub
|
cbe2baf596 |
feat(web): add skills & design systems management page in settings (#535)
* feat(web): add skills & design systems management page in settings Add a new "Library" section in Settings that lets users browse, search, preview, and enable/disable skills and design systems. Disabled items are excluded from the create-project picker. Phase 1 — browse/toggle only. Closes #497 * fix(web): persist empty disabled lists and deduplicate DS preview Use empty array instead of undefined when all items are re-enabled so the daemon merge clears the key. Move DS preview panel outside the category group loop so it renders once, not per group. * fix(web): address review feedback on library settings Clear disabled lists on invalid daemon writes, memoize enabled item filters in App.tsx, and guard preview fetch against rapid-click race conditions. * fix(web): hydrate disabled lists from daemon and keep full lists in ProjectView Merge daemonConfig.disabledSkills/disabledDesignSystems during bootstrap so the values survive localStorage resets. Pass unfiltered skills and design systems to ProjectView so existing project metadata resolves correctly. |
||
|
|
cc6da191e8 |
fix(version): resolve daemon package.json from any compiled layout (#537)
Settings -> About used to display 0.0.0 in packaged builds because `readCurrentAppVersionInfo` resolved `'../package.json'` relative to `import.meta.url`, which only points at the daemon package root from the flat CLI build (`dist/app-version.js`). The sidecar build emits `dist/src/app-version.js`, where the same relative path lands on the non-existent `dist/package.json`, so `readPackageMetadata` returned null and the version fell back to APP_VERSION_FALLBACK. Walk up from `import.meta.url` to find the nearest real `package.json` instead, so the daemon reports its actual version regardless of whether it runs from TypeScript source (tools-dev), the flat CLI dist, or the nested sidecar dist used by the packaged desktop app. The OD_APP_VERSION env still wins inside `resolveAppVersionInfo`, so callers that already inject it (mac/win packagers) keep working. Also write `appVersion` into the Linux packaged config so Linux follows the same env-injection path as mac/win and stays consistent with the new fallback resolution. Co-authored-by: Cursor <cursoragent@cursor.com> |
||
|
|
79fcaef129 |
Add Tweaks mode for HTML previews with picker, pod selection, and batched chat attachments (#513)
* Add tweaks mode for HTML preview comments * Fix tweaks geometry and restore critique migration * Harden tweaks mode reload sync * Guard tweaks batch sends during active runs --------- Co-authored-by: puma <puma@pumas-MacBook-Air.local> |
||
|
VioletandGitHub
|
34e8db175d |
fix(daemon): preserve ANTHROPIC_API_KEY when ANTHROPIC_BASE_URL is set (#514)
* fix(daemon): preserve ANTHROPIC_API_KEY when ANTHROPIC_BASE_URL is set The claude adapter currently strips ANTHROPIC_API_KEY unconditionally so that Claude Code's own auth resolution (claude login) wins instead of silently falling back to API-key billing. However, when ANTHROPIC_BASE_URL is set the user is intentionally routing Claude Code to a custom endpoint (e.g. a Kimi/Moonshot proxy). In that case claude login is meaningless, so preserve the API key so the child can authenticate against the custom base URL. * fix(daemon): guard against empty ANTHROPIC_BASE_URL values Address review feedback: check that ANTHROPIC_BASE_URL contains a non-empty, non-whitespace string before preserving ANTHROPIC_API_KEY. This prevents the #398 billing guard from being bypassed when the variable is set to '' or whitespace. |
||
|
|
e0a75ea8c8 |
fix: add lowercase 'todowrite' compatibility in ToolCard (#523)
Add backward compatibility for historic OpenCode CLI data: - Match both 'TodoWrite' and 'todowrite' tool names - Fixes rendering of TodoWrite cards from persisted chat history Scope: 1 line change, no other files modified Co-authored-by: bojiehuang <bojiehuang@bojiehuangdeMacBook-Pro.local> |
||
|
Marc ChanandGitHub
|
c3d9136a0c |
Add live artifacts and Composio connector catalog (#381)
* docs: add live artifacts implementation spec * docs: align live artifacts implementation plan * Ralph iteration 1: work in progress * Ralph iteration 2: work in progress * Ralph iteration 3: work in progress * Ralph iteration 4: work in progress * Ralph iteration 5: work in progress * Ralph iteration 6: work in progress * Ralph iteration 7: work in progress * Ralph iteration 8: work in progress * Ralph iteration 9: work in progress * Ralph iteration 10: work in progress * Ralph iteration 11: work in progress * Ralph iteration 12: work in progress * Ralph iteration 13: work in progress * Ralph iteration 14: work in progress * Ralph iteration 15: work in progress * Ralph iteration 16: work in progress * Ralph iteration 17: work in progress * Ralph iteration 18: work in progress * Ralph iteration 19: work in progress * Ralph iteration 20: work in progress * Ralph iteration 21: work in progress * Ralph iteration 22: work in progress * Ralph iteration 23: work in progress * Ralph iteration 24: work in progress * Ralph iteration 25: work in progress * Ralph iteration 26: work in progress * Ralph iteration 27: work in progress * Ralph iteration 28: work in progress * Ralph iteration 29: work in progress * Ralph iteration 30: work in progress * Ralph iteration 31: work in progress * Ralph iteration 32: work in progress * Ralph iteration 33: work in progress * Ralph iteration 34: work in progress * Ralph iteration 35: work in progress * Ralph iteration 36: work in progress * Ralph iteration 37: work in progress * Ralph iteration 38: work in progress * Ralph iteration 39: work in progress * Ralph iteration 40: work in progress * Ralph iteration 41: work in progress * Ralph iteration 42: work in progress * Ralph iteration 43: work in progress * Ralph iteration 44: work in progress * Ralph iteration 45: work in progress * Ralph iteration 46: work in progress * Ralph iteration 47: work in progress * Ralph iteration 48: work in progress * Ralph iteration 49: work in progress * Ralph iteration 50: work in progress * Ralph iteration 51: work in progress * Ralph iteration 52: work in progress * Ralph iteration 53: work in progress * Ralph iteration 54: work in progress * Ralph iteration 55: work in progress * Ralph iteration 56: work in progress * Ralph iteration 57: work in progress * Ralph iteration 58: work in progress * Ralph iteration 59: work in progress * Ralph iteration 60: work in progress * Ralph iteration 61: work in progress * Ralph iteration 62: work in progress * Ralph iteration 63: work in progress * Ralph iteration 64: work in progress * Ralph iteration 65: work in progress * Ralph iteration 1: work in progress * Ralph iteration 2: work in progress * Ralph iteration 3: work in progress * Ralph iteration 4: work in progress * Ralph iteration 5: work in progress * Ralph iteration 6: work in progress * Ralph iteration 8: work in progress * Ralph iteration 9: work in progress * Ralph iteration 17: work in progress * Add Composio-backed connectors * Add Composio-backed connector catalog * Fix connector callback flow * Update live artifact connector refresh * Fix live artifact refresh updates * Improve live artifact viewer toolbar * Refine live artifact source tabs * Expand Composio connector catalog * Improve Composio connector browsing * Fix artifact refresh source safety checks Generated-By: looper 0.4.1 (runner=fixer, agent=opencode) * Fix live artifacts PR feedback Generated-By: looper 0.5.0 (runner=fixer, agent=opencode) * Fix live artifact preview CORS validation Generated-By: looper 0.0.0-dev (runner=fixer, agent=opencode) * Fix connector OAuth IPv6 loopback hosts Allow bracketed IPv6 loopback Host headers when deriving connector OAuth callback URLs so IPv6-bound daemons can complete connection flow. Generated-By: looper 0.0.0-dev (runner=fixer, agent=opencode) * Preserve live artifact refresh permissions Respect explicit refresh permission choices during live artifact create and update flows so revoked connector sources remain gated. Generated-By: looper 0.0.0-dev (runner=fixer, agent=opencode) * Fix live artifact preview cache freshness Generated-By: looper 0.0.0-dev (runner=fixer, agent=opencode) * Fix live artifact refresh validation Guard manual refreshes with local daemon checks and reject daemon_tool sources without a toolName before refresh execution. Generated-By: looper 0.0.0-dev (runner=fixer, agent=opencode) * Fix Composio credential invalidation Generated-By: looper 0.0.0-dev (runner=fixer, agent=opencode) * Fix live artifact CORS methods Generated-By: looper 0.0.0-dev (runner=fixer, agent=opencode) * Fix workspace validation Restore media config test isolation under Vitest setup data-dir overrides and add the missing French live artifact display copy so the workspace test suite stays aligned.\n\nGenerated-By: looper 0.5.2 (runner=fixer, agent=opencode) * Fix connector safety filtering Keep agent-preview connector listings aligned with execution safety policy and prune stale Composio OAuth state records before they accumulate. Generated-By: looper 0.5.2 (runner=fixer, agent=opencode) * Fix agent runtime cleanup Generated-By: looper 0.5.2 (runner=fixer, agent=opencode) * Fix live artifact daemon access Validate local-only live artifact routes against the peer socket address and pass daemon-resolved CLI paths to ACP MCP descriptors.\n\nGenerated-By: looper 0.5.2 (runner=fixer, agent=opencode) * Fix connector run limit pruning Evict stale connector rate-limit buckets so long-lived daemon processes do not retain per-run entries indefinitely.\n\nGenerated-By: looper 0.5.2 (runner=fixer, agent=opencode) * Fix connector compact schemas Generated-By: looper 0.5.2 (runner=fixer, agent=opencode) * Improve connector connection feedback * Adjust connector gate positioning * Fix live artifact refresh commits Avoid marking refresh candidates failed after snapshot or state persistence errors by deferring live artifact mutations until the durable refresh metadata is written. Also align connector OAuth callback host validation with daemon loopback handling.\n\nGenerated-By: looper 0.5.4 (runner=fixer, agent=opencode) * Improve connector search relevance * fix(daemon): harden connector connection state Require loopback daemon validation before connector connect side effects and only clear provider-owned connector statuses during credential reset. Generated-By: looper 0.5.4 (runner=fixer, agent=opencode) * fix(daemon): guard connector disconnect route Require local daemon request validation before connector disconnect side effects. Generated-By: looper 0.5.4 (runner=fixer, agent=opencode) * fix(daemon): guard composio config updates Generated-By: looper 0.5.4 (runner=fixer, agent=opencode) * fix(daemon): dispatch live artifacts mcp first Route the live-artifacts MCP server before the generic MCP CLI so od mcp live-artifacts starts the dedicated server instead of failing generic argument parsing.\n\nGenerated-By: looper 0.5.4 (runner=fixer, agent=opencode) * fix(daemon): handle integer connector schemas Allow JSON Schema integer connector inputs while preserving fractional-value validation so generated connector tool schemas accept valid page sizes and limits. Generated-By: looper 0.5.4 (runner=fixer, agent=opencode) * fix: align live artifact refresh error codes Generated-By: looper 0.5.4 (runner=fixer, agent=opencode) * Fix live artifact connector refresh flow * Update live artifact design cards * Add beta badge to live artifact form * Remove live artifact tile model * Fix live artifact refresh sync * Fix live artifact MCP refresh durability Generated-By: looper 0.5.4 (runner=fixer, agent=opencode) * Fix live artifact refresh safety Enforce persisted refresh opt-out and connector auto-read gating before refresh sources execute. Generated-By: looper 0.5.5 (runner=fixer, agent=opencode) |
||
|
|
76e6c7a9f6 |
feat: Critique Theater Phase 4 (persistence + transcript + orchestrator) (#481)
* docs(specs): add Critique Theater design spec for panel-tempered artifacts * docs(specs): add Critique Theater implementation plan * docs(specs): rename UI to Design Jury, add lane-density modes, ship-rule explainer, label sizing * feat(contracts): add CritiqueConfig schema and defaults * fix(contracts): apply Task 1.1 review (CRITIQUE_PROTOCOL_VERSION rename, descriptions, RoleWeights export) * feat(contracts): add PanelEvent discriminated union and isPanelEvent guard * fix(contracts): apply Task 1.2 review (exhaustive event-type list, runId guard, import order) * feat(contracts): add CritiqueSseEvent variants and panelEventToSse mapper * test(daemon): add v1 wire-protocol golden fixtures for Critique Theater parser * feat(daemon): add v1 streaming parser for Critique Theater wire protocol * chore(contracts): add .js extensions to relative imports for NodeNext consumers * fix(daemon): satisfy noUncheckedIndexedAccess in v1 parser regex match access * test(daemon): cover parser failure modes; fix unclosed-PANELIST swallow bug * fix(daemon,contracts): address PR #387 review - parser now clamps panelist + DIM scores against the run-declared scale captured from <CRITIQUE_RUN scale=...>, not a hardcoded 100 - PANELIST appearing before any <ROUND n=...> opens now throws MalformedBlockError rather than emitting events with NaN round - DIM_RE and MUST_FIX_RE hoisted to module scope and lastIndex reset per call so the parser hot path stops recompiling regex per artifact - overflow check after drain simplified to a plain buf.length > cap test (the prior compound condition was always true on the right side and obscured intent) - scoreThreshold <= scoreScale refine gains a 1e-9 epsilon so floating slack does not reject semantically valid configs - round-1 designer ARTIFACT guard gains a comment naming the spec invariant and the v2 relaxation path - 3 new regression tests cover the panelist-without-round, scale=10 clamp, and scale=20 plumbing cases * docs(specs): rationale for non-goals, failure-mode rate targets, Phase 10 matrix, Phase 14 doc layout * Merge branch 'main' into feat/critique-theater Resolves the contracts/index.ts conflict by keeping the .js extensions added by chore(contracts) |
||
|
PerishFireandGitHub
|
bbdd4e84b5 |
chore: enforce test directory conventions (#496)
* chore: enforce test directory conventions Move package, app, and tool tests out of src and add guard enforcement so source directories stay source-only. * ci: use guard and package-scoped tests Run the new repository guard in CI and keep test execution aligned with package-scoped commands after removing root aliases. * ci: align stable release guard check Use the new repository guard in stable release verification after replacing the residual-JS-only script. * chore: tighten test layout enforcement Enforce sibling tests directories, typecheck moved test suites with dedicated configs, and refresh remaining guidance that pointed at src-based tests. * chore: clarify no-emit test tsconfigs Explicitly disable declaration-only emit in test tsconfigs so review tooling sees they are no-emit typecheck configs. |
||
|
Chris TamandGitHub
|
a3a0ae6ced |
fix(daemon): respect baseUrl path verbatim in OpenAI-compat proxy (#410)
* fix(daemon): respect baseUrl path verbatim in OpenAI-compat proxy `appendVersionedApiPath` previously force-injected `/v1` unless the supplied baseUrl ended with `/vN`. That broke any provider whose OpenAI-compatible surface lives under a sub-path: https://api.deepinfra.com/v1/openai → ".../v1/openai/v1/chat/completions" https://openrouter.ai/api/v1 → ".../api/v1/chat/completions" (worked, by luck of /vN suffix) Now the auto-`/v1` only fires when the user supplied no path at all, so DeepInfra, OpenRouter, and any other sub-path-mounted compat surface route to the right endpoint while the canonical `https://api.openai.com` / `https://api.anthropic.com` shortcuts still work. Adds a regression test table covering the matrix. * Account for Anthropic style URLs |
||
|
GoemonandGitHub
|
b5598572d8 |
fix(daemon): scope -32603 recovery to session/set_model request id (#492)
Address review feedback from lefarcen and mrcfps on PR #492: P1: Track setModelRequestId to scope the recovery block to the exact session/set_model request. This prevents duplicate prompt sends if session/prompt returns -32603 (which would otherwise match on expectedId + non-default model conditions). P1: Add promptRequestId === null guard so the recovery path only triggers before a prompt has been sent. P2: In detectAcpModels, only suppress -32603 on unexpected ids. Expected-id -32603 errors (initialize, session/new) are real probe failures and should reject immediately rather than causing a silent 15s timeout. P2: In attachAcpSession, expected-id -32603 errors that don't match setModelRequestId now call fail() instead of falling through. This prevents initialize/session/new/session/prompt failures from being silently swallowed. |
||
|
|
0ab98fed07 |
fix(daemon): correct Claude Code --add-dir capability detection (fix #430) (#440)
Probe 'claude -p --help' (subcommand help) instead of 'claude --help' (global help), because --add-dir and --include-partial-messages are subcommand-specific flags that do not appear in the global help output. Previously, successful capability probing would set caps.addDir = false because the flag was not found in global help, causing the daemon to never pass --add-dir to spawned Claude Code processes. This broke absolute-path reads for skill seeds and design-system specs, as Claude Code's sandbox policy blocks access to paths outside the project cwd unless explicitly allowed via --add-dir. The probing-failure / timeout path already worked correctly (empty caps object defaults to passing --add-dir). The failure manifested only when help probing succeeded and explicitly failed to locate the subcommand flags. Co-authored-by: bojiehuang <bojiehuang@bojiehuangdeMacBook-Pro.local> |
||
|
Justin GaoandGitHub
|
cc8add4f09 |
feat(daemon): add link code folder support for agent context (#455)
* feat(daemon): add link code folder support for agent context Users can now link local code directories to a project so the AI agent reads their source code via --add-dir when generating designs. The import menu's "Link code folder" item opens a native OS folder picker, and linked folders appear as removable chips below the chat input. - Add linkedDirs field to ProjectMetadata contract - Add POST /api/dialog/open-folder endpoint (osascript/zenity/PowerShell) - Add validateLinkedDirs with path safety checks (absolute, exists, blocklist) - Append linked dirs to extraAllowedDirs in startChatRun - Add system prompt hint listing linked code folders - Render linked folder chips in ChatComposer with add/remove - Add i18n strings for all 16 locales - Add 8 unit tests for validateLinkedDirs * fix: address PR review feedback - Add JSDoc type annotations to validateLinkedDirs for strict mode - Check path.isAbsolute before resolve to catch relative inputs - Allow linking when projectMetadata is undefined (default to prototype) - Remove redundant PATCH in ProjectView callback * fix: use inline TS annotations instead of JSDoc in linked-dirs.ts * fix(daemon): harden linked-dirs validation against security bypasses - Resolve symlinks with realpathSync.native before checking blocklist - Reject filesystem root (/) and drive roots as linked dirs - Canonicalize blocklist entries to handle macOS /etc -> /private/etc - Validate linkedDirs on project creation, not just PATCH - Re-validate persisted linkedDirs in startChatRun before use - Add tests for root, symlink-to-blocked-dir, and realpath resolution * fix: narrow union type before accessing .error in tests |
||
|
PerishFireandGitHub
|
3935aeb421 |
Optimize packaged mac artifact size (#424)
* optimize mac package payload reporting * optimize(pack): package standalone web runtime * optimize(pack): default to standalone web runtime * chore(release): bump beta base version * fix(pack): compress mac artifacts and report packaged version * fix(pack): preserve Next server fallback * fix(pack): clarify standalone startup failures * fix(release): gate beta platform builds * fix(web): bind standalone backend to parent * fix(pack): harden standalone and beta publishing |
||
|
aitonicandGitHub
|
e5ed4bbc76 |
fix(packaged): preserve *_API_KEY env vars for CLI agents (#404)
* Forward API key env vars to packaged sidecars ## Summary Forward provider credential environment variables from the packaged desktop app process to packaged sidecar children. ## Why The packaged app currently forwards only a small fixed environment allowlist to sidecars: - HOME - LANG - LC_ALL - LOGNAME - TMPDIR - USER This prevents CLI agents launched by the packaged daemon from seeing provider credentials such as `CODEX_API_KEY`, even when the desktop app itself was launched with that variable set. ## Fix This keeps the existing allowlist and additionally forwards environment variables ending in: - `_API_KEY` - `_TOKEN` I intentionally did not change PATH or Bun-related behavior in this PR, since CLI discovery already works. The issue here is specifically credential forwarding. ## Verification I verified the root cause locally by patching the packaged output: `@open-design/packaged/dist/index.mjs` After forwarding `CODEX_API_KEY`, the packaged desktop app was able to launch Codex successfully. ## Testing I did not run the full project build locally. I verified the behavior by patching the installed macOS app output and confirming that forwarding `CODEX_API_KEY` fixes the Codex launch failure. * Add includeProviderSecrets parameter to env functions Updated. Provider-secret forwarding is now daemon-only: *_API_KEY / *_TOKEN are included only when spawning the daemon sidecar, while the web/Next sidecar stays on the original minimal allowlist. |
||
|
|
9979afa3ae |
fix: cap htmlPreviewSlideState Map to prevent unbounded memory growth (#488)
The module-level Map cached slide state per project+file key but never evicted entries. In long-lived sessions with many projects, this grows without bound. Add a 64-entry cap with FIFO eviction. Co-authored-by: Test User <test@example.com> Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com> |
||
|
Ajay SatishandGitHub
|
ec7dafc007 |
feat(daemon): add Kilo CLI (ACP) (#480)
* docs(readme): refresh contributors wall * docs(readme): refresh contributors wall * feat: kilo cli * fix: use default model option for kilo * chore: add agent_diff id unique test * chore: add deepseek to docs |
||
|
IrixandGitHub
|
6380c48a48 | Update Codex sandbox invocation (#477) | ||
|
ChildhoodAndyandGitHub
|
de62baa0e8 |
fix(web): aspect ratio cards no longer overflow into siblings (#476)
The 380px sidebar gives the 3-column aspect grid only ~92px per cell. The flex-row card layout reserved 34px for the glyph plus 7px gap, leaving ~31px for the label — but "Landscape"/"Portrait" at 13px / weight 650 are ~75px wide. Without `min-width: 0` on the flex/grid items, the strong text refused to shrink and overflowed, so each card's label visually covered the next card's glyph. Switch the aspect cards to a column layout (glyph above, label + ratio below) so the label has the full card width. Add the missing `min-width: 0` on `.newproj-option-card` and `.aspect-copy` as a defense against future long-text regressions, and rebalance glyph dimensions and font sizes to match the tighter cards. Closes #242 |
||
|
jiakebogeandGitHub
|
2473ab9567 |
fix(web): add copy buttons for FileViewer code blocks (#471)
* fix(web): add copy buttons for FileViewer code blocks * fix(web): harden FileViewer markdown copy controls * fix(web): restore focus after clipboard fallback * test(e2e): restore execCommand after markdown copy tests |
||
|
ChildhoodAndyandGitHub
|
a4fb90a9f3 |
fix(design-files): clear selection on project switch (#465)
* fix(design-files): clear selection on project switch to prevent cross-project leak
When switching from project A to project B, filenames present in both
projects (e.g. index.html) would survive the intersection-prune and
remain selected, causing handleBatchDownload() to post against the
wrong projectId. Now selection is fully cleared on projectId change.
* fix(design-files): use key-based remount to prevent cross-project selection leak
Replace the useEffect projectId clearing with a key={projectId} on
DesignFilesPanel. useEffect runs after render, leaving a window where
stale selections from project A could be used against project B. The
key tells React to unmount and remount the component synchronously on
project switch — no async gap.
|
||
|
alburtandGitHub
|
b4e69ac61b | fix(copilot): copilot prompt processing with correct command format (#466) | ||
|
ChildhoodAndyandGitHub
|
bc04d61903 |
feat(design-files): add batch ZIP download with multi-select (#405)
* feat(design-files): add batch ZIP download with multi-select Add checkbox-based multi-select to the Design Files panel and a new POST /api/projects/:id/archive/batch endpoint that zips selected files using the project name as the archive filename. * fix(i18n): add missing batch-download keys to uk locale The upstream main branch has a uk.ts locale that didn't exist in the fork. Without these keys the web typecheck fails against the full locale set. * fix(design-files): harden batch archive and prune stale selections - Use lstat to reject symlinks, dotfiles, and .artifact.json sidecars in buildBatchArchive (mirror listFiles/collectFiles allowlist) - Reject invalid names explicitly instead of silently skipping - Prune stale filenames from selected set on files/projectId change * fix(daemon): tighten batch archive allowlist with segment-level checks - Check every path segment for hidden directories, not just basename - Walk intermediate directories with lstat to reject symlinks at any level - Fail-fast on any ineligible file instead of silently skipping |
||
|
Tom HuangandGitHub
|
da2b007a43 |
feat(daemon): add DeepSeek TUI as a code agent adapter (#439)
* feat(daemon): add DeepSeek TUI as a code agent adapter
Register `deepseek` (with `deepseek-tui` cargo-only fallback) in
AGENT_DEFS via `deepseek exec --auto [--model X] <prompt>` and plain-text
streaming. Ships `deepseek-v4-pro` / `deepseek-v4-flash` as fallback
model hints; users can paste any other id (incl. NIM / Fireworks /
SGLang routes) via the custom-model input.
Web UI gets a DeepSeek-blue gradient icon, label/alias mapping, and
docs/agent-adapters.md §5.9 documents the auth state, prompt-as-argv
Windows size limit, and the upstream gap that prevents stdin delivery
today (clap declares `prompt: String` as a required positional).
Adds .deepseek/ to .gitignore alongside the other per-agent runtime
data dirs so first-launch trust files don't leak into git.
* fix(daemon): drop unsupported deepseek-tui fallback bin
The `deepseek` dispatcher owns `exec` / `--auto`; `deepseek-tui` is the
runtime companion it invokes. Listing `deepseek-tui` in fallbackBins
advertised availability for a host that only had the TUI binary, but
buildArgs still emitted `<resolved> exec --auto <prompt>` — which
deepseek-tui itself doesn't accept, so the first /api/chat run would
fail. Upstream documents both binaries as required (npm and cargo paths
install them together), so the fallback didn't correspond to a supported
install. Pin the absence in the agents test and update docs §5.9 + the
adapter table to match.
Generated-By: looper 0.4.0 (runner=fixer, agent=claude-code)
* fix(daemon): pre-flight DeepSeek TUI prompts against argv byte budget
DeepSeek's exec mode requires the prompt as a positional argv arg (no
`-` stdin sentinel upstream), so a fully composed OD prompt — system
text + history + skills + design-system content + the user message —
can blow Windows' ~32 KB CreateProcess limit (or Linux MAX_ARG_STRLEN
on extreme edges) and surface as a generic spawn failure instead of
a DeepSeek-specific, user-actionable message. The adapter now declares
`maxPromptArgBytes = 30_000` (leaves ~2.7 KB argv headroom for `exec
--auto --model <id>` and Windows quoting), and the /api/chat spawn
path checks the composed prompt against that budget before calling
`spawn`. Oversized prompts fail fast with `AGENT_PROMPT_TOO_LARGE`
and guidance to reduce skills/design context or pick an adapter with
stdin support.
Generated-By: looper 0.4.0 (runner=fixer, agent=claude-code)
* test(daemon): pin DeepSeek argv-budget guard with regression tests
The previous spawn-path guard inlined the byte-budget check in the
chat handler, so the only safety net for the DeepSeek argv-only
prompt-delivery shape was a static "the field exists" assertion —
nothing actually exercised the AGENT_PROMPT_TOO_LARGE path or the
short-prompt happy path. Extract the check into a pure
`checkPromptArgvBudget(def, composed)` helper in agents.ts, call it
from /api/chat before bin resolution (so the guard is order-
independent and fires regardless of whether the adapter binary is
on PATH in CI), and add a regression test that exercises both the
oversized-prompt branch (over the conservative under-Windows-
CreateProcess budget) and the short-prompt branch, plus a UTF-8
byte-vs-codepoint case and a stdin-adapter no-op case so the guard
can't silently regress or leak onto adapters that ship the prompt
over stdin.
Generated-By: looper 0.4.0 (runner=fixer, agent=claude-code)
* fix(daemon): pre-flight DeepSeek prompts against Windows .cmd-shim quoting
The first-pass argv-byte guard only inspects the raw composed prompt, so
on Windows an npm-installed `deepseek` resolves to a `.cmd` shim and the
spawn path then wraps the call in `cmd.exe /d /s /c "<inner>"` with
every embedded `"` doubled by `quoteWindowsCommandArg`. A quote-heavy
prompt (code blocks, JSON-shaped skill seeds) under the 30,000-byte
budget can therefore still expand past CreateProcess's 32_767-char
`lpCommandLine` cap and surface as a generic spawn ENAMETOOLONG instead
of the DeepSeek-named, actionable `AGENT_PROMPT_TOO_LARGE` the budget
guard was meant to provide. Add a second pure helper
`checkWindowsCmdShimCommandLineBudget(def, resolvedBin, args)` that
mirrors the platform layer's per-arg quoting and recomputes the would-be
command line length whenever the resolved binary is a `.cmd` / `.bat`
shim, and call it from `/api/chat` after `buildArgs` / `resolveAgentBin`
so the same SSE error fires before `spawn`. Pin the new path with a
quote-heavy regression (prompt is under the byte budget but doubles
past the kernel cap) plus no-op tests for non-`.cmd` resolutions, null
bin, and stdin-only adapters so the guard can't drift back.
Generated-By: looper 0.4.0 (runner=fixer, agent=claude-code)
* fix(daemon): extend DeepSeek argv guard to direct .exe Windows installs
The cmd-shim guard added in
|
||
|
nmsnandGitHub
|
59e1e01233 |
fix(web): prevent vertical scrollbar on artifact preview frame (#453)
* fix(web): prevent vertical scrollbar on artifact preview frame The preview uses transform: scale() for zoom, which doesn't affect layout size. When previewScale < 1, the scale wrapper's layout size (100/scale%) overflows the parent, triggering overflow: auto on .viewer-body. Add overflow: hidden on .comment-preview-layer to clip the layout overflow before it propagates. * fix(web): use iframe-only clip wrapper instead of layer-level overflow Move overflow: hidden from .comment-preview-layer to a dedicated .comment-frame-clip wrapper that only wraps the iframe. This prevents the layout overflow from propagating to .viewer-body while keeping comment overlays and popovers unclipped. |
||
|
emilneanderandGitHub
|
33c3b94b42 |
feat(daemon): add od mcp - expose Open Design as an MCP server (#399)
* feat(daemon): add `od mcp` subcommand for stdio MCP server
Lets a coding agent in a different repo (Claude Code, Cursor, Zed)
pull files from a locally-running OD project over the Model Context
Protocol — no export/import zip dance.
The MCP server is a thin stdio process that proxies read-only tool
calls to the daemon's existing HTTP API; no daemon-side changes
required. Exposes 8 tools:
list_projects, get_project,
list_files, get_file,
list_skills, get_skill,
list_design_systems, get_design_system
Wired exactly like `od media`: a hoisted flag set, a SUBCOMMAND_MAP
entry, a thin handler that resolves OD_DAEMON_URL and hands off to
src/mcp.ts. Tool dispatch is a switch over the tool name; each branch
fetches the matching daemon route and surfaces the response as MCP
text content. Binary mimes return a clear error pending phase-2
support.
Lifecycle gotcha worth flagging: Server.connect(transport) only
*starts* the stdio reader; the promise resolves immediately. Without
holding the function awaiting until transport/stdin close, cli.ts's
top-level process.exit(0) kills the server before the first request
arrives. The fix in src/mcp.ts holds until onclose / stdin EOF.
Wire-up example for a consuming repo:
{
"mcpServers": {
"open-design": {
"command": "od",
"args": ["mcp"],
"env": { "OD_DAEMON_URL": "http://127.0.0.1:7456" }
}
}
}
New dep: @modelcontextprotocol/sdk (MIT, official Anthropic SDK).
* feat(daemon): add MCP server instructions for zero-shot LLM context
Hand the consuming LLM a system-prompt-style overview of the OD
workflow so it picks the right tool without prompt-engineering on
the user's side. Mentions get_artifact and project-name resolution
ahead of their actual implementation; both ship in the same batch.
* feat(daemon): resolve MCP project args by UUID, name, or substring
Lets a consuming agent say `project: "recaptr"` instead of pasting a
UUID. Match order: exact id → exact name (case-insensitive) →
slug-normalized name (strips trailing " (N)", normalizes whitespace) →
substring (errors if multiple). UUID inputs short-circuit and never
hit the daemon.
* feat(daemon): surface entryFile and kind on MCP get_project response
Promote metadata.entryFile and metadata.kind to top-level fields so
consumers (including get_artifact in this branch) can find the entry
without digging through nested metadata blobs.
* feat(daemon): add MCP get_artifact tool for bundle retrieval
A design rarely lives in a single file. get_artifact pulls the entry
HTML/JSX plus every sibling it references (tokens CSS, JSX modules,
imported components) in one call, so a consuming agent doesn't need
to parse HTML and round-trip per file.
Three modes:
auto (default): BFS over relative <script src>, <link href>,
<img src>, <source/video src>, JSX import/from, CSS url(), with
depth cap 3 and a visited set. CDN, data:, mailto:, anchors, and
paths containing .. are skipped.
all: every textual file in the project (mirror of /archive
minus binaries).
shallow: just the entry file (same as get_file).
Output is a structured JSON blob with name/mime/size/content per
file and the project's manifest metadata at the top.
* feat(daemon): add /api/projects/:id/search route + MCP search_files
Server-side substring search across textual project files. Returns
file, 1-indexed line, and snippet, capped at 1000 matches. Exposed
through the MCP layer as search_files(project, query, pattern?, max?).
Treats the query as a literal substring (regex chars escaped) to
avoid catastrophic-backtracking attacks from LLM-supplied input.
Honors the project dir's existing path-safety guards via listFiles.
* feat(daemon): add since= filter to /files route + MCP list_files arg
Lets a consumer poll for "what's changed since I last looked" without
re-walking every file. Daemon-side: parse since= as ms, filter
listFiles output by mtime. MCP-side: forward as URL query.
* feat(daemon): expose skills and design systems as MCP resources
Catalog reads are stable reference material — they fit MCP's
resources surface (LLM-passive) better than tools (LLM-active).
Skills and design systems each become resources at
od://skills/<id>/SKILL.md and od://design-systems/<id>/DESIGN.md;
existing list_skills / get_skill / list_design_systems /
get_design_system tools remain as fallbacks for clients that don't
handle resources cleanly.
* fix(daemon): tighten MCP correctness in get_artifact and resources
Several silent-failure paths and minor footguns the first pass missed:
- get_artifact auto: the entry's own fetch now raises a clear
error instead of returning files: []. Previously a typo in
`entry:` looked like an empty project.
- get_artifact: invalid `include` value returns a clear error
listing the valid modes instead of silently behaving as auto.
- get_artifact all: includes binary files as metadata stubs to
match auto's behavior. Both modes are now strict supersets of
shallow.
- extractRelativeRefs: gate JS-only patterns (import/from/require/
dynamic-import) by file mime/extension so prose in markdown or
HTML doesn't generate spurious 404 round-trips on words like
"imported from 'X'".
- extractRelativeRefs: cover <iframe>, <audio>, srcset, and
CSS @import — common in real OD output.
- resources/list descriptions are collapsed to a single line
(newlines + repeated whitespace -> one space) so MCP UIs that
don't normalize whitespace render cleanly.
- fetchProjectFile: 0-byte binary files no longer report size: null
due to falsy short-circuit on Number(content-length).
* perf(daemon): cache MCP project list for 5s in resolveProjectId
A typical agent session calls list_files/get_file/get_artifact several
times in a row, each with a project name argument. Each previously
re-fetched /api/projects. Cache the list in module scope with a 5s
TTL so back-to-back lookups are local; renames in the OD UI still
propagate within a few seconds.
* feat(daemon): MCP UX polish — tool order, annotations, get_artifact maxBytes
Three changes well-behaved MCP clients pick up automatically:
- Tool ordering. list_projects + get_artifact are now first; LLMs
that weight earlier entries surface the bundle path before
per-file fetching. Catalog tools (list_skills, get_skill,
list_design_systems, get_design_system) sit at the bottom; they
are also exposed as MCP resources.
- readOnlyHint / idempotentHint / openWorldHint annotations on
every tool so clients can skip confirmation prompts on safe
tools and let the LLM know re-running is fine. Per-tool `title`
annotations give clients a friendlier display name than the
snake_case tool id.
- get_artifact gains a `maxBytes` arg (default 1.5MB). Once the
accumulated textual content crosses the cap, remaining files
are dropped and `truncated: true` is set on the bundle so the
consumer knows to use list_files / get_file for the rest.
* feat(daemon): expose user's active OD project/file via MCP
The "what file are you on?" round-trip the agent had to do every
session is now answered automatically. Three pieces:
- Daemon: in-memory active-context slot with 5-minute TTL.
POST /api/active sets {projectId, fileName}; GET /api/active
returns the current value enriched with projectName, or
{active:false} when the slot is empty/stale. Cleared on
daemon restart.
- Web: a small useEffect in App.tsx posts the active project +
file to the daemon on every route change. Best-effort fire-
and-forget; a missing daemon doesn't surface an error.
- MCP: get_active_context tool (no args) and a matching MCP
resource at od://focus/active. The tool is listed second,
right after list_projects, so an LLM picks it up before
asking for ids. Server instructions tell the model to call
it FIRST when the user says "this file" / "the design I have
open" / "what I'm looking at."
End to end: user opens a project in OD, agent in another repo
calls get_active_context() → gets {projectName: "recaptr",
fileName: "recaptr-onboarding-4.html"}, then immediately calls
get_artifact(project: "recaptr") with no further user input.
* feat(daemon): make MCP project arg optional, fall back to active OD context
get_artifact, get_project, get_file, search_files, and list_files now
accept project as optional. When omitted, the MCP resolves project
from /api/active so an agent in another repo can call
search_files({ query: "Polaroid" })
without first asking the user "which project?". get_file and
get_artifact also default their path/entry to the active file, so
get_file({}) returns whatever the user is currently looking at.
The implicit path stamps `usedActiveContext` on JSON responses (or a
separate `[od:active-context …]` content block on get_file) so the
agent can see exactly which project/file got chosen. Explicit
project args pass through with zero added overhead.
Cuts the common case from two MCP round trips
(get_active_context → search_files) to one. Server instructions and
get_active_context's own description are updated to point at the
new default.
* fix(daemon): require same-origin for /api/active POST and GET
The active-context endpoint was added without isLocalSameOrigin
guard. Since the daemon binds 0.0.0.0 by default, a LAN peer could
GET it to learn what file the user has open, or POST it to redirect
the MCP fallback to a project of their choice. Same-origin only is
the right scope: the web app proxies its requests through Next.js
on the daemon port, and the MCP runs over loopback in-process, so
both legitimate callers pass.
Pattern matches the existing /api/app-config etc. guards.
* feat(daemon): add /api/mcp/install-info for cross-platform install snippets
The Settings -> MCP server panel needs absolute paths to node and
the daemon's built cli.js so it can render snippets that work on a
fresh source clone (where `od` is not on PATH) and dodge the
/usr/bin/od octal-dump tool that ships on macOS/Linux and would
otherwise shadow ours.
Endpoint returns:
- command: process.execPath (the node binary running the daemon)
- args: [<absolute path to dist/cli.js>, "mcp"]
- daemonUrl: http://127.0.0.1:<port>
- platform: process.platform (so the panel can localize ~/.cursor
vs %USERPROFILE%\.cursor and Cmd vs Ctrl shortcuts)
- cliExists / nodeExists: existsSync checks on both binaries
- buildHint: human-readable build/reinstall instructions when
either path is missing
isLocalSameOrigin guard same as /api/active. Cached for 5s because
the panel may re-fetch on every open and the paths cannot change
without a daemon restart.
Test file covers the happy path, cross-origin rejection, two
allowed-Origin variants, and the cache by counting fresh resolves
across rapid calls. 5/5 pass.
* refactor(daemon): tighten MCP surface, trim descriptions, polish copy
Three intertwined cleanups that all live in mcp.ts + cli.ts:
1. Drop catalog tools from MCP. list_skills / get_skill /
list_design_systems / get_design_system are removed. The audience
is a coding agent in a separate repo consuming Open Design's
output; it cannot run skills (those are recipes Open Design uses
to generate) and design-system DESIGN.md is reference material
that already ships as an MCP resource. Keeping the catalog as
tools cost ~350 token-overhead per turn for capabilities the
agent could not act on. Tool count: 11 -> 7.
2. Trim tool descriptions. The active-context fallback explanation
was repeated in 5 separate tool descriptions; hoisted into
PROJECT_ARG and explained once in the server `instructions`
block instead. Saves ~150-200 tokens per tools/list response.
3. User-facing branding pass. Tool titles, tool descriptions,
resource names, error messages, comments, and `od mcp --help`
now consistently use "Open Design" rather than "OD". Internal
abbreviation `OD` is retained only inside the server
instructions block where it is introduced inline as "Open Design
(OD)" for compactness across multi-paragraph guidance.
Em dashes replaced with hyphens throughout, per project style.
* feat(web): add MCP server install panel in Settings
New "MCP server" section in the Settings dialog, surfacing
copy-paste install snippets for the major MCP-compatible coding
agents (Claude Code, Cursor, VS Code, Antigravity, Zed, Windsurf).
Highlights:
- In-brand custom dropdown (reuses the existing .ds-picker
pattern from the design-system / prompt-template pickers, click
outside / Escape to close, chevron animates) instead of a
native <select>.
- Per-client snippet that uses absolute paths to node + cli.js
fetched from /api/mcp/install-info on mount, so it works even
when `od` is not on PATH.
- Cursor gets a one-click "Install in Cursor" deeplink
(cursor://anysphere.cursor-deeplink/mcp/install) that pops an
approval dialog and writes the config for the user. UTF-8-safe
base64 so paths with accented characters do not throw.
- Per-OS path hints (~/.cursor on POSIX, %USERPROFILE%\.cursor
on Windows) and keyboard shortcuts (Cmd vs Ctrl).
- Build-required warning card when cli.js or the node binary
does not exist on disk; deeplink button disables in that state.
- Prominent "restart your client to pick up the new server"
callout below the snippet, with per-client guidance.
- Capability list ("what your agent can do") instead of a tool-
name dump, so non-developer designers can also tell what is
possible without reading MCP docs.
README adds a short "Use Open Design from your coding agent"
section that points at the panel and summarizes the per-client
flow (one-click for Cursor, JSON merge elsewhere). Read-only by
design; the daemon must be running locally.
* docs(readme): align MCP server section with the Settings panel
The "Use Open Design from your coding agent" section had drifted
from what the panel actually emits and lists.
- Add Antigravity to the supported-client list (previously missing).
- Drop the "(GitHub Copilot)" parenthetical from VS Code so the
label matches the panel.
- Fix the Claude Code line: we no longer emit a single
`claude mcp add ...` shell command. The snippet is JSON; the
panel additionally suggests `claude mcp add-json` as the safer
way to apply it instead of hand-editing ~/.claude.json.
- Swap the "find the Polaroid section" example for two more
universal phrases ("build this in my app", "match these
styles") that match what the panel surfaces.
- Add a one-line "restart or reload your client after install"
note - this was prominent in the panel and absent from the
README.
- Trim the /usr/bin/od octal-dump aside; it was technical detail
that did not earn its space at the README intro level.
* feat(web): add Codex CLI to the MCP server install panel
Codex is a first-class supported coding agent (listed alongside
Claude Code, Cursor, etc. in the README's PATH-detected agent
table) but the install panel was missing it.
Codex stores MCP server config at ~/.codex/config.toml (TOML, not
JSON) under an `[mcp_servers.<name>]` table, and the same file is
shared between the Codex CLI and the Codex IDE extension - so one
install covers both. Added a 7th client entry that emits the right
TOML snippet, expanded the snippet-lang union to include 'toml'
(behaves like 'json' for whitespace handling, just a different
syntax-highlight hint).
For our minimal payload (just command + args), JSON.stringify
happens to produce valid TOML literal values since TOML basic
strings use the same double-quote escape rules as JSON, and TOML
inline arrays match JSON array syntax. No new TOML serializer
needed.
README updated to list Codex among the supported clients.
Schema verified against https://developers.openai.com/codex/mcp.
* fix(daemon): accept any loopback origin in same-origin guard
The previous port-pinned check required the request's Origin to match
either the daemon's own port or OD_WEB_PORT. tools-dev does not pass
OD_WEB_PORT to the daemon process, so any browser POST to /api/active
proxied through the dev web (port 17573 etc.) was rejected with 403,
and get_active_context always returned {active: false}.
Relax to a loopback-prefix match: any http://127.0.0.1:*,
http://localhost:*, or http://[::1]:* origin passes regardless of
port. Cross-origin (https://evil.com) is still rejected. The
trade-off is that another local web app on a different loopback port
could now CSRF the daemon; same-origin checks are inherently a CSRF
defense, not a network ACL.
* fix(web): make Claude Code MCP snippet a real copyable one-liner
claude mcp add-json open-design '<json>' takes only the inner
server-config object, not the full {"mcpServers": ...} wrapper, and
rejected the wrapped shape with "Invalid configuration: : Invalid
input". Pass only the inner config, and inline the JSON into the
command itself so the snippet is a real one-liner the user can copy
and paste, no template substitution.
* test(daemon): drop loopback-prefix assertions superseded by upstream origin policy
The two proxy-flow allow tests were added in
|
||
|
nmsnandGitHub
|
02d24c4fde |
fix(web): language option button height truncation in settings (#447)
* fix(web): remove min-height constraint on language option buttons The hardcoded min-height: 40px on .settings-language-option caused content to overflow when locale labels wrap to two lines. Remove the constraint so buttons size naturally, and update the menu max-height budget from 360px (9×40px) to 406px (7×58px) to match. * fix(web): bump menu max-height to 428px to account for chrome The previous 406px budget didn't account for menu padding (4px), row gaps (2px × 6), and border (1px × 2). Bump to 428px so all 7 visible items fit without clipping. |
||
|
Daniel Duma, PhDandGitHub
|
cfd359e05a |
[codex] Fix Gemini CLI trust handling (#352)
* Fix Gemini CLI trust handling * Preserve agent spawn env filtering |
||
|
nmsnandGitHub
|
e9cc1fbda8 |
fix(web): prevent vertical scrollbar on ws-tabs-bar (#448)
Setting overflow-x: auto implicitly sets overflow-y to auto per CSS spec. With a fixed height of 44px, this causes a vertical scrollbar when content overflows by even 1px. Explicitly set overflow-y: hidden. |