Files
buzz/scripts/run-differential.mjs
T
cc00060ea3 feat(agent): Phase 2 — wire Rust and TS consumers to generated model-capabilities module (#3958)
## What

Phase 2 consumer cutover targeting the
`duncan/databricks-model-label-registry` umbrella branch. Wires
`crates/**` and `desktop/**` consumers to the generated capability
module introduced in Phase 1 (#3821), while keeping old and new paths
both live for differential testing. Phase 3 removes the old paths.

## Commits (boundary-separated)

### feat(agent): Phase 2a — wire Rust consumers to generated capability
module (`crates/**`, `scripts/**`)

- `catalog.rs`: `DATABRICKS_V2_KNOWN_MODELS` re-exported from the
generated module — single source of truth.
- `llm.rs`: `databricks_v2_route_for_model` delegates to
`resolve_model_capabilities("databricks_v2", model)`. Old segment-based
classifier preserved as `#[cfg(test)] _old_*` for the differential
harness. New `databricks_v2_route_differential_old_vs_new` test confirms
100% agreement on all 20 route vectors.
- `config.rs`: new `effort_table_fixture_differential_old_vs_new` test
runs `resolve_model_capabilities` over the 36-entry
`effortTable.fixture.json` and asserts old/new agree modulo a doc-cited
allowlist (4 F1 corrections).
- `scripts/run-differential.mjs`: JS differential harness over
effortTable fixture + normative corpus + catalog-sample fixture. 85
checks, 0 unexpected divergences (5 allowlisted: 4 F1 corrections +
goose-opus-5 anthropic route correction).
- `scripts/MODELS_DEV_RECONCILIATION.md`: deferred MINOR from Phase 1 —
8 trailing-double-space line breaks replaced with `<br>`.

### feat(desktop): Phase 2b — cut TS consumers to generated
model-capabilities module (`desktop/**`)

- `buzzAgentConfig.ts`: adds
`getProviderEffortConfigFromManifest(provider, model?)` — thin wrapper
over `resolveModelCapabilities()` from `modelCapabilities.ts`. Maps
`supportedEfforts → validValues` and `defaultEffort → defaultValue`
(null preserved for manual-budget/Inherit). Old
`getProviderEffortConfig()` and all hand-tables stay live for the
differential harness; Phase 3 retires them.
- `formatAgentModelLabel.ts`: registry-label lookup re-pointed from
hand-maintained `databricksModelNames.ts` import to generated
`DATABRICKS_MODEL_NAMES` exported from `modelCapabilities.ts`. Same Map
shape, identical contents, behavior unchanged.

### fix(scripts): add ts-esm-loader and fix allowlist coverage in
run-differential (`scripts/**`)

- `scripts/ts-esm-loader.mjs`: minimal ESM custom loader that resolves
extensionless relative TS imports. Required because Phase 2b's
`buzzAgentConfig.ts` imports `modelCapabilities` without `.ts` extension
— which Node's `--experimental-strip-types` runner cannot resolve
without a hook.
- `scripts/run-differential.mjs`: shebang updated to self-bootstrap with
the loader; fixes the `totalAllowlisted` counter (was declared but never
incremented — always printed `0 allowlisted`). Replaced with per-axis
hit tracking: reports exercised slot count (`N/total`) in summary; fails
with `STALE_ALLOWLIST` if any declared entry fires zero divergences,
preventing stale entries from silently masking future regressions.

## Verification

- `cargo test -p buzz-agent --lib`: 426/426
- Corpus: 45/45 · schema-negative: 24/24 · `--check` byte-clean
- Differential: 85 checks, 0 unexpected divergences, 6/6 allowlist slots
exercised
- Desktop: 3847/3847 · typecheck clean · biome clean
- Mobile: 1019 pass, 1 skipped — same 5 flaky tests in
`mobile/test/features/channels/` that reproduce at the umbrella base;
zero mobile files in this branch range
- `git diff --check`: clean

## What Remains (Phase 3)

Remove old hand-maintained paths: `_old_*` functions in
`llm.rs`/`config.rs`, old `getProviderEffortConfig` tables in
`buzzAgentConfig.ts`, old `databricksModelNames.ts` import in
`formatAgentModelLabel.ts`, old `databricks_model_names.rs` module.

---------

Signed-off-by: Will Pfleger <pfleger.will@gmail.com>
Signed-off-by: npub1g8493u0xfsjrvflg4n08ezd7vec99mnwzlv0qgwpr9d7gvjwhuzqx59rhw <41ea58f1e64c243627e8acde7c89be667052ee6e17d8f021c1195be4324ebf04@buzz.block.builderlab.xyz>
Signed-off-by: npub1mn7jgtj4w2pd0g0zeuhxsa6jy6p0rewxz4kujt98my82ahfmp72sxjexk7 <dcfd242e557282d7a1e2cf2e6877522682f1e5c6156dc92ca7d90eaedd3b0f95@buzz.block.builderlab.xyz>
Co-authored-by: npub1mn7jgtj4w2pd0g0zeuhxsa6jy6p0rewxz4kujt98my82ahfmp72sxjexk7 <dcfd242e557282d7a1e2cf2e6877522682f1e5c6156dc92ca7d90eaedd3b0f95@buzz.block.builderlab.xyz>
Co-authored-by: npub1g8493u0xfsjrvflg4n08ezd7vec99mnwzlv0qgwpr9d7gvjwhuzqx59rhw <41ea58f1e64c243627e8acde7c89be667052ee6e17d8f021c1195be4324ebf04@buzz.block.builderlab.xyz>
2026-08-03 14:58:34 -04:00

240 lines
8.3 KiB
JavaScript
Executable File

#!/usr/bin/env node
/**
* Phase-2 differential harness — compare old buzzAgentConfig.ts effort logic with
* the new generated modelCapabilities.ts interpreter over:
* 1. The 36-entry effortTable.fixture.json (cross-boundary Rust/TS fixture)
* 2. The 45-vector normative corpus (scripts/normative-corpus.json)
* 3. The catalog-sample fixture (scripts/catalog-sample-fixture.json)
*
* Equality is required except for entries in the committed allowlist of intentional
* F1 corrections (models.dev provider-capability reconciliations).
*
* Usage: node --experimental-strip-types scripts/run-differential.mjs [--verbose]
* Exits 0 on all-pass (modulo allowlist), 1 on unexpected divergence or unexercised allowlist entry.
*/
import { readFileSync } from "node:fs";
import { join, dirname } from "node:path";
import { fileURLToPath } from "node:url";
const __dirname = dirname(fileURLToPath(import.meta.url));
const repoRoot = join(__dirname, "..");
const VERBOSE = process.argv.includes("--verbose");
// ---------------------------------------------------------------------------
// Import both interpreters
// ---------------------------------------------------------------------------
// NEW: generated capability module
const { resolveModelCapabilities: resolveNew } = await import(
join(repoRoot, "desktop", "src", "features", "agents", "ui", "modelCapabilities.ts")
);
// OLD: buzzAgentConfig.ts effort config
const { getProviderEffortConfig_oldHandTable: getOldEffortConfig } = await import(
join(repoRoot, "desktop", "src", "features", "agents", "ui", "buzzAgentConfig.ts")
);
// ---------------------------------------------------------------------------
// Intentional corrections allowlist (Phase 1 F1 reconciliations)
// Each entry: { provider, raw_model_id, reason }
// ---------------------------------------------------------------------------
const ALLOWLIST = [
{
provider: "databricks_v2",
raw_model_id: "databricks-gpt-5-5",
axes: ["supported_efforts"],
reason: "Phase 1 ADOPT: models.dev d5a4974c advertises [low,medium,high]; old returns [none,low,medium,high,xhigh]",
},
{
provider: "databricks_v2",
raw_model_id: "databricks-gpt-5-4-mini",
axes: ["supported_efforts"],
reason: "Phase 1 ADOPT: models.dev advertises [low,medium,high]; old returns [none,low,medium,high,xhigh]",
},
{
provider: "databricks_v2",
raw_model_id: "databricks-gpt-5-4-nano",
axes: ["supported_efforts"],
reason: "Phase 1 ADOPT: models.dev advertises [low,medium,high]; old returns [none,low,medium,high,xhigh]",
},
{
provider: "databricks_v2",
raw_model_id: "databricks-gpt-5-6-sol",
axes: ["supported_efforts"],
reason: "Phase 1 ADOPT: models.dev advertises [low,medium,high,max]; old returns [none,low,medium,high,xhigh,max]",
},
{
provider: "databricks_v2",
raw_model_id: "goose-opus-5",
axes: ["supported_efforts", "default_effort"],
reason: "Phase 1 correction: 'opus' is a named DBv2 segment → anthropic-messages route; old config.rs disagreed with llm.rs (corpus note dbv2-goose-opus-5-is-anthropic). Generated adopts anthropic adaptive-xhigh capabilities consistent with the wire route.",
},
];
// Track which allowlist entries are actually exercised (suppressed a divergence).
// Keyed as "provider:raw_model_id:axis".
const allowlistHits = new Set();
function isAllowlisted(provider, rawModelId, axis) {
const entry = ALLOWLIST.find(
(e) =>
e.provider === provider &&
e.raw_model_id === rawModelId &&
e.axes.includes(axis),
);
if (entry) {
allowlistHits.add(`${provider}:${rawModelId}:${axis}`);
return true;
}
return false;
}
// ---------------------------------------------------------------------------
// Comparison helpers
// ---------------------------------------------------------------------------
/**
* Compare effort axes from both interpreters for one (provider, model) pair.
* Returns array of divergence objects.
*/
function compareEffortAxes(provider, model) {
const newResult = resolveNew(provider, model);
const oldResult = getOldEffortConfig(provider, model);
const divergences = [];
// supported_efforts
const newEfforts = newResult.supportedEfforts ?? [];
const oldEfforts = oldResult?.validValues ?? [];
if (JSON.stringify(newEfforts) !== JSON.stringify(oldEfforts)) {
if (!isAllowlisted(provider, model, "supported_efforts")) {
divergences.push({
axis: "supported_efforts",
old: oldEfforts,
new: newEfforts,
});
}
}
// default_effort
const newDefault = newResult.defaultEffort ?? null;
const oldDefault = oldResult?.defaultValue ?? null;
if (newDefault !== oldDefault) {
if (!isAllowlisted(provider, model, "default_effort")) {
divergences.push({
axis: "default_effort",
old: oldDefault,
new: newDefault,
});
}
}
return divergences;
}
// ---------------------------------------------------------------------------
// Test suites
// ---------------------------------------------------------------------------
let totalChecks = 0;
let totalDivergences = 0;
function runCheck(label, provider, model) {
totalChecks++;
const divs = compareEffortAxes(provider, model);
if (divs.length > 0) {
totalDivergences += divs.length;
for (const d of divs) {
console.error(
`DIVERGE [${label}] provider=${provider} model=${model} axis=${d.axis}\n` +
` old: ${JSON.stringify(d.old)}\n` +
` new: ${JSON.stringify(d.new)}`,
);
}
} else if (VERBOSE) {
console.log(`OK [${label}] provider=${provider} model=${model}`);
}
}
// 1. effortTable.fixture.json
console.log("--- effortTable.fixture.json ---");
const fixture = JSON.parse(
readFileSync(
join(repoRoot, "desktop", "src", "features", "agents", "ui", "effortTable.fixture.json"),
"utf8",
),
);
for (const entry of fixture) {
if (!entry.provider) continue;
runCheck("fixture", entry.provider, entry.model ?? "");
}
// 2. normative-corpus.json (effort axes only)
console.log("--- normative-corpus.json ---");
const corpus = JSON.parse(
readFileSync(join(repoRoot, "scripts", "normative-corpus.json"), "utf8"),
);
for (const entry of corpus) {
if (entry._group) continue;
if (!entry.provider || !entry.expect) continue;
if (!entry.expect.supported_efforts && !entry.expect.default_effort) continue;
runCheck("corpus", entry.provider, entry.raw_model_id ?? "");
}
// 3. catalog-sample-fixture.json (exact records from pinned models.dev payload)
console.log("--- catalog-sample-fixture.json ---");
const catalogFixture = JSON.parse(
readFileSync(join(repoRoot, "scripts", "catalog-sample-fixture.json"), "utf8"),
);
for (const ep of catalogFixture.endpoints ?? []) {
if (!ep.name) continue;
// All catalog endpoints are databricks_v2 provider
runCheck("catalog-sample", "databricks_v2", ep.name);
}
// ---------------------------------------------------------------------------
// Summary
// ---------------------------------------------------------------------------
// Count total allowlist axis slots expected to be hit
const totalAllowlistSlots = ALLOWLIST.reduce((n, e) => n + e.axes.length, 0);
const allowlistHitCount = allowlistHits.size;
// Detect stale allowlist entries (declared but never actually suppressed a divergence)
const staleEntries = [];
for (const entry of ALLOWLIST) {
for (const axis of entry.axes) {
const key = `${entry.provider}:${entry.raw_model_id}:${axis}`;
if (!allowlistHits.has(key)) {
staleEntries.push({ ...entry, axis });
}
}
}
console.log(
`\nDifferential: ${totalChecks} checks, ${totalDivergences} unexpected divergences, ${allowlistHitCount}/${totalAllowlistSlots} allowlist slots exercised`,
);
if (staleEntries.length > 0) {
for (const e of staleEntries) {
console.error(
`STALE_ALLOWLIST provider=${e.provider} model=${e.raw_model_id} axis=${e.axis} — entry never fired; remove or update it`,
);
}
}
if (totalDivergences > 0) {
console.error(
`FAIL: ${totalDivergences} unexpected divergence(s) — see output above`,
);
process.exit(1);
} else if (staleEntries.length > 0) {
console.error(
`FAIL: ${staleEntries.length} stale allowlist entry(ies) — entries that never suppress a divergence mask future regressions`,
);
process.exit(1);
} else {
console.log("PASS: old and new effort logic agree on all non-allowlisted entries");
}