fix(aif-handoff-bridge): Stage 1→2 Phase -1 follow-up — B1+M1+m1/m2/m3 - #269
Merged
Merged
Conversation
Mechanical corrections to merged SW-A (#268) + SW-B (#267) research-patches per Phase -1 Stage 1→2 cold-review. No verdict changes — SW-A REFERENCE and SW-B REJECT both unaffected. - B1: SW-B :39, :117 — PR #128 IS merged. Original record used jq camelCase '.mergedAt' which returns null; the actual snake_case '.merged_at' returns 2026-05-26T08:07:12Z (merge SHA 51ce96a). Verified 2026-05-29 via 'gh api repos/lee-to/aif-handoff/pulls/128 --jq .merged_at'. - M1: SW-A :343 — gate-4 evidence said "all are from PR #128" but BOTH PR #127 (e1ffa70, merged 2026-05-15) AND PR #128 (51ce96a, merged 2026-05-26) are in the 30-day window. Both touch packages/runtime/ only. - m1: SW-A :392 — Probe 6 URL hash typo ae7472a9 -> ae7072a9 (aligned to line 119 body-text canonical version; inferred from internal consistency, not externally re-verified due to DeepWiki client-side rendering). - m2: SW-B :210 — overclaim "≥4 distinct phrasings on dir-watch" corrected to actual breakdown: 2 direct + 2 adjacent DeepWiki + 1 file-tree + 1 gh + 3 WebSearch = 9 channels. Multi-channel triangulation noted. - m3: SW-B §0 falsifier (:17) — aligned to §5 specificity ("…and creates tasks via the internal createTask function — re-evaluate as ADAPT with three criteria 1/2/4 re-scored…"). Phase -1 Stage 1→2 cold-review verdict was REVISE; this patch closes B1+M1+m1/m2/m3 dispositions per state.md §3.4. DN-1 and DN-2 remain maintainer-call; SW-C dispatch still blocked on those answers. Prior-art: skipped — mechanical text correction, no capability commit, no new dependency, no new file ≥50 LOC. Per CLAUDE.md capability-commit definition, this is a refactor/typo-fix grade change to existing artifacts.
artyhoo
enabled auto-merge (squash)
May 29, 2026 09:56
artyhoo
added a commit
that referenced
this pull request
May 29, 2026
… Implementer-equivalent only) value-add audit (#276) R-phase patch for Sub-wave C of the aif-handoff-as-runtime-bridge umbrella. Evaluates Variant C (kickoff §3 lines 124-145): aif-handoff as Implementer- equivalent only, bypass Planner+Reviewer cycle, thin CLI wrapper for kickoff dispatch + kanban status tracking. Verdict: REJECT (BFR-default §1 ladder). Rationale: - The kickoff-framed "aif-handoff exec --kickoff <path>" CLI does not exist in lee-to/aif-handoff (DeepWiki probes 1+5, 2026-05-29). - No first-class Implementer-only mode; skipReview:true bypasses Reviewer but Planner is mandatory unless accept_existing_plan with on-disk PLAN.md (same disk coupling SW-A flagged for Variant A). - BEFORE/AFTER maintainer-action count: 25% literal / 0% cognitive reduction (T-AIF-BRIDGE-C table §4) — below kickoff §8 STOP 30% threshold → verdict "Variant C value-add insufficient". - Pure-tracker pattern (paused:true + autoMode:false + manual state-machine transitions) IS shipped but adds zero automation beyond UI tracking; Docker+SQLite infra unjustified. Cites: - SW-A merged PR #268 (Variant A REFERENCE, 28% match, 3 ADOPT-blockers) - SW-B merged PR #267 (Variant B REJECT, ~5% match, no dir-watch capability) - PR #269 follow-up (mechanical corrections, no verdict changes) - DN-1=B-constrained input consumed in criterion 5 (mooted for Variant C which bypasses aif-handoff Reviewer entirely) - Gate-4 admission re-sweep: PR #127+#128 touch packages/runtime/ only (no MCP/coordinator drift in 30-day window) 5 distinct DeepWiki probes + 2 WebSearches + cross-ref to SW-A/SW-B/PR #269 = 19+ evidence channels (T1 floor exceeded 3.8x). §1.7 forward+backward + §self-application + T-trap walk per ai-laziness-traps.md §3. Single output file under docs/meta-factory/research-patches/. No code, skill, agent, install.sh, or .claude/rules/ modifications. ### §1.7 Forward-check applied build-first-reuse-default.md §1 verdict ladder applied; BFR §3 6-layer search performed (SSOT rows #27/#28/#29/#30/#43/#44/#46/#67/#80 reviewed at prior-art-evaluations.md:95-148; DeepWiki >=5 probes; WebSearch >=2 phrasings; own-stack sweep at .claude/skills/meta-orchestrator/SKILL.md:441 anti-scope + :404+:429 SP requesting-code-review). no-paid-llm-in-ci.md §1 enforced (all evidence via subscription-bundled DeepWiki/WebSearch + free gh CLI + bash). reviewer-discipline.md §2 respected (DN-1=B-constrained consumed as fact, not re-litigated; verdict is research finding against §8 STOP, not strategy choice). ai-laziness-traps.md §3 active T-traps applied (T1, T3, T7, T11, T12, T13, T15, T16, T17, T19, T20, T-AIF-BRIDGE-C MANDATORY BEFORE/AFTER table at patch §4). Evidence: see patch §8 file:line citations. ### §1.7 Backward-check applied SSOT #27/#28/#67 receive additive notes (additive-only; no verdict changes). Original DEFER/DEFER/REJECT rationales reviewed at prior-art-evaluations.md: 95, 96, 135 — consistent with Sub-wave C findings (reinforce existing classifications, do not re-litigate). No .claude/rules/* modified; no .claude/skills/* modified; no agents/* modified; no packages/* modified; no install.sh modified; no kickoff.md modified. Single output file in docs/meta-factory/research-patches/. Scope strictly bounded to Variant C; SW-A/SW-B/SW-B2/SW-D out of scope. T15 self-application confirmed in patch §10. Memory not written (Sub-wave D synthesis is the natural codification surface). Evidence: see patch §9 file:line citations.
6 tasks
artyhoo
added a commit
that referenced
this pull request
May 29, 2026
…DONE (#283) Per umbrella's post-GO directive (Phase -1 Stage 3 close-review GO 2026-05-29 on PR #281 SW-D synthesis), update wave-sequencing-plan.md §0 to reflect umbrella closure: - Row update: stale 🟡 «Sub-wave A + B R-phase done / synthesis pending» → ✅ DONE 2026-05-29 with verdict DEFER ALL. Full per-variant scoring (A REFERENCE 28% / B REJECT ~5% / B' REFERENCE 35% conditional / C REJECT 22%) + 3 ADOPT-blockers shared across A/B'/C + 6 evidence PRs (#267/#268/#269/ #275/#276/#281) + SSOT additive notes status (proposed, not landed). - Renamed row to canonical umbrella name (`aif-handoff-as-runtime-bridge`). - «What actually remains» — remove the umbrella entry; add to deferred-decisions parenthetical at bottom of the line. - Plan revision history — add this PR. No verdict changes to any other §0 row. No edits outside §0. Broader 40+ UNTRACKED PRs in §0 (recordkeeping debt per state.md Decision 2) stay deferred — out of scope for this umbrella's closure per CLAUDE.md PR strategy. Prior-art: skipped — recordkeeping update for completed umbrella, no capability commit, no new dependency, no new file ≥50 LOC. Per CLAUDE.md capability-commit definition, this is a docs-edit grade change.
12 tasks
artyhoo
added a commit
that referenced
this pull request
Sep 5, 2026
…yer (#1597 ledger A6-* R-6 R-7 S-4) (#1625) ## Summary Nine findings from the #1597 promote review's runtime-bridge CLI share (A6-1 A6-2 A6-4 A6-6 A6-7 A5-7/A6-5 R-6 R-7 S-4, plus the D-4 src twin of A6-4), fixed in `packages/runtime-bridge/src/cli/` and re-vendored into `packages/runtime-bridge/vendor/` — the copy a factory-profile install places at `.claude/vendor/runtime-bridge/`, i.e. the code a consumer actually runs. Includes the routed port of #1618's four src fixes to the vendored twins, which that PR left behind (C13 → C12 handoff). The through-line is one defect class: the same helper hand-copied per CLI, diverging silently. Four copies of a main-module guard (one form correct since #968, four naive, one file with no guard at all), five copies of an argv lookup, three copies of an HTTP request + BackendError mapping. Each is now one module the others import. ## Changes - **New `src/cli/cliEntry.ts`** — shared entrypoint plumbing. - `isMain()` realpaths BOTH sides, so a CLI invoked through a symlink still runs (A6-1 / R-6). Applied to all eight CLIs ([park.ts:227](packages/runtime-bridge/src/cli/park.ts), answer:269, questions:228, ensure-parallel:135, dispatch:322, claim:190, harvest:689, await:154). - `parseCliArgs()` wraps `node:util` `parseArgs` (built-in, no dependency) and replaces five hand-rolled `argv.indexOf(flag) + 1` lookups; rejects unknown flags, a flag whose value is the next flag, a flag value hijacking a positional, empty `--flag=`, and positional junk (A6-4 / A6-7 / D-4 / R-6). - **`dispatch.ts`** exits 1, not 0, when the kickoff path is absent or unreadable (A6-2). The PostToolUse hook is unaffected — it reads stdout in a command substitution and exits 0 unconditionally. - **`park.ts`** is idempotent: a repeat park of the same question issues no PUT and reports `alreadyParked` (A6-6). - **`harvest.ts`** `main()`: `getTask` and the `--body-file` read moved inside the try that owns the `[harvest] FAILED:` line and the Channel-A fallback (A5-7 / A6-5); the bare top-level `void main()` is now guarded. - **`cli/aifHttp.ts`** is the single request implementation: `answer.ts`'s verbatim copy (S-4) and `aifWsStatus.getTaskStatus`'s third hand-written `GET /tasks/:id` (R-7) both call it. One timeout policy and one 429 → `quota_exceeded` mapping, where the copies had already diverged. - **Vendored twins** re-synced (`cp` + `scripts/format-shipped.sh --write`, P4 still parked), and #1618's four src fixes ported to them in their own commit. ## Prior-art consult - [x] Capability commits in this PR carry a `Prior-art:` trailer citing SSOT entries by ID. - [x] A new capability area surfaced (argv parsing + ESM main-module detection for a dependency-free vendored CLI tree): entries **#268** (`node:util parseArgs` — ADOPT; `commander`/`yargs`/`minimist` REJECTED in the same consult) and **#269** (`es-main` — ADAPT: its realpath-both-sides mechanism, not the dependency) added to [prior-art-evaluations.md](docs/meta-factory/prior-art-evaluations.md) in the same commit as `cliEntry.ts`, each with Verdict / Rationale / Trigger to revisit. - [x] No existing entry matched the area (searched the §4 table for CLI/argv/entrypoint capability rows). - [x] context7 (`/nodejs/node`, `util.parseArgs` strict mode / allowPositionals / missing-value errors) plus two WebSearch phrasings (ESM main-module detection under symlinks; `parseArgs` vs `minimist`/`commander`/`yargs`), all 2026-09-05, cited in entry #268/#269 rationales. ## Test plan - [x] `npm test` in `packages/runtime-bridge` — **337 passed, 1 skipped** (was 317 before; 20 new assertions). - [x] `npx tsc --noEmit` in `packages/runtime-bridge` — clean; **and the vendored tree typechecks standalone** with `vendor/` un-excluded (it is outside the framework's `tsc` scope by design). - [x] `npm run format:check` (shipped surface, includes the vendor drop) — clean. - [x] `bash scripts/build-getff-dist.sh --check` — in sync, 1060 files. - [x] `bash tests/install-sh/gh-531-shipped-prettier.test.sh` — PASS=48 FAIL=0. - [x] `bash tests/hooks/runtime-bridge-dispatch-optin.test.sh` — PASS=5 FAIL=0. - [x] `packages/core` — `hooks/runtime-bridge-dispatch.test.ts` + `hooks/checks/prior-art.test.ts` — 126 passed. - [x] Manual smoke: each of the four naive-guard CLIs driven through a real symlink (`test/cli-symlink-entry.test.ts`) — RED before this change with empty stderr and exit 0. ## Provenance n/a ## Review findings Fixed here (each verified by a test that was RED on the pre-change code): - **A6-1 — four CLIs silently no-op under a symlinked path.** Failure-scenario: an agent runs `npx tsx /tmp/<worktree>/…/park.ts --task X --question '…'` where the absolute path crosses a symlink (macOS `/tmp` → `/private/tmp`, a `bin/` shim, `node_modules/.bin`); the guard compares a resolved path against an unresolved one, is false, and the process exits 0 with no stdout and no stderr — the task is never paused, and `questions.ts` reads the same way as "no parked questions". - **A6-2 — dispatch exits 0 on an unreadable kickoff.** Failure-scenario: `/dispatcher` or a night loop runs `dispatch.ts` with a mistyped kickoff path; stderr carries the ENOENT but `$?` is 0 and stdout is empty, so a wrapper checking the exit code records the stage as dispatched while no task exists. - **A6-4 / D-4 — a flag value before the task id becomes the task id.** Failure-scenario: `harvest.ts --base staging f1010da4` harvests task `staging` → `GET /tasks/staging` 404s outside the try → raw stack; the real task is never harvested. - **A6-7 — a flag whose value is the next flag.** Failure-scenario: `park.ts --task --question 'Fork…'` yields `taskId='--question'`, passes validation, and `GET /tasks/--question` 404s as `dispatch_failed` instead of reporting the missing `--task`. - **A6-6 — park is not idempotent.** Failure-scenario: an agent retry or night-loop re-entry runs `park.ts --task X --question Q` a second time; the plan gains a second identical OPEN QUESTION block and `blockedReason` is overwritten, so the operator sees one fork twice. - **A5-7 / A6-5 — harvest failures bypass the classified exit path.** Failure-scenario: an unreachable aif, an unknown taskId, or a mistyped `--body-file` throws before the try opens, so the operator gets a raw unhandled-rejection stack with no `[harvest] FAILED:` line and no Channel-A fallback commands — exactly when they need the manual egress. - **R-6 — the guard and the argv lookup copied 4×/5×.** Failure-scenario: the realpath fix from #968 reached two of six CLIs, and harvest's argv copy had additionally dropped the truthiness guard (`--base ""` → empty base branch); a fix in one copy does not reach the others by construction. - **R-7 — `getTaskStatus` re-implements `getTask` + its mapping.** Failure-scenario: the copies had already diverged — the status probe aborts after 5 s while the CLI copy had no timeout at all, so `park`/`answer`/`harvest`/`ensure-parallel` hang forever on a wedged aif API that the probe of the same URL times out against; and `getTaskStatus` had no 429 branch, reporting a rate-limited API as `dispatch_failed`. - **S-4 — `answer.ts post()` duplicates `aifHttp request()`.** Failure-scenario: the aif error-mapping contract lives in two functions per tree and the tree is vendored, so four copies; adding a timeout or changing 429 handling must be replayed per copy or they diverge (which R-7 shows already happened). Ported from #1618 (routed by C13, not re-derived here): unscoped `GET /runtime-profiles`, missing-`projectId` classification, the unguarded cancel DELETE, and the silent `exit(0)` on a non-BackendError — all four still live in the vendored twins at that PR's merge. Verification that the port is a port: prettier@3.8.3 over the HEAD `src` of all ten vendored modules reproduces the committed vendor byte-for-byte for seven and differs for exactly the three #1618 touched. Not fixed here, surfaced for routing: - **A6-6 residue** — parking a `done`/`verified`/`backlog` task still sets `paused:true` on a terminal task, and `answer.ts` `resumePark` still rewrites plan/paused on a task that is not paused. Narrowed deliberately: refusing a terminal-status park changes the `/pipeline` `done` → `request_changes` flow, which is outside this PR's scope. - **`AifHandoffBackend._rest`** ([AifHandoffBackend.ts:530](packages/runtime-bridge/src/AifHandoffBackend.ts)) is a fourth copy of the same request + mapping, left untouched: backend semantics are C13's ownership zone and #1618 was in flight on that file. - **E-1** — `vendor/src/idempotency.ts` still hardcodes the `/tmp` dedup path the vendor README documents as `RUNTIME_BRIDGE_DEDUP_PATH`-driven. Consumer-fit item, C14's zone. ## Fidelity verdict FIDELITY: skipped — no kickoff: this is a review-ledger fix batch dispatched from the #1597 finding list, not a stage of an umbrella ## Parked questions none ## §1.7 Self-discipline check (REQUIRED if PR touches discipline-bearing files) ### §1.7 Forward-check applied This PR touches [docs/meta-factory/prior-art-evaluations.md](docs/meta-factory/prior-art-evaluations.md) (two new rows), which is discipline-bearing, so the Forward+Backward pair applies rather than the Skipped line. Disciplines checked before adding them: - **Build-vs-reuse / own-stack-first** — [.claude/rules/build-first-reuse-default.md:46](.claude/rules/build-first-reuse-default.md) §1.1. The capability (argv parsing) was resolved by a Node built-in and an existing in-repo implementation, not by a new dependency and not by a new module of our own invention; both candidate families are recorded with verdicts rather than silently skipped. - **SSOT entry procedure** — [prior-art-evaluations.md:51](docs/meta-factory/prior-art-evaluations.md) §3: entries added in the SAME commit as the capability artifact (`cliEntry.ts`), each carrying Verdict / Rationale / Trigger to revisit, with the rationale citing external links (nodejs/node#57616) and measured evidence rather than «didn't fit». - **`#sync-by-copy-paste`** — [.claude/rules/dual-implementation-discipline.md:199](.claude/rules/dual-implementation-discipline.md) §8. The defect class this PR fixes is exactly that anti-pattern; the counter applied is the one the rule names — extract the shared implementation, make one channel the SSOT and have the others import it. - **Attention is not a mechanism** — [.claude/rules/attention-is-not-a-mechanism.md:17](.claude/rules/attention-is-not-a-mechanism.md) §1. Every finding lands as a deterministic test that was RED on the pre-change code, not as a comment warning future readers; the symlink case in particular is driven through a real symlink rather than asserted about. - **Capability-commit definition** — [CLAUDE.md](CLAUDE.md) «What is a capability commit?»: `cliEntry.ts` is a new ≥80 LOC file under `packages/`, so it takes the SSOT-citation path, not the escape hatch (the pre-push gate rejected the escape hatch and that rejection is what produced entries #268/#269). ### §1.7 Backward-check applied Sweep of the existing artefacts under the same scope, before and after: - **Every CLI in the tree, not only the four the ledger named.** Enumerated all of `packages/runtime-bridge/src/cli/*.ts` for the guard defect: four naive (park, answer, questions, ensure-parallel), two already correct since #968 ([dispatch.ts:322](packages/runtime-bridge/src/cli/dispatch.ts), [claim.ts:190](packages/runtime-bridge/src/cli/claim.ts)), and two the ledger did not report — [harvest.ts:689](packages/runtime-bridge/src/cli/harvest.ts) had a bare top-level `void main()` and [await.ts:154](packages/runtime-bridge/src/cli/await.ts) had no guard at all, both found by the import side-effect their absence produced. All eight now share the one helper. - **Every copy of the argv lookup**: five sites (park, answer, questions, ensure-parallel, harvest), all replaced; `await.ts`'s own `parseArgs` is a different shape (`--timeout-ms` numeric coercion) and was left, noted here rather than silently. - **Every copy of the request + BackendError mapping**: three folded into [aifHttp.ts](packages/runtime-bridge/src/cli/aifHttp.ts) (`answer.post`, `aifWsStatus.getTaskStatus`, plus `getTask`/`putTask` which already lived there); the fourth, [AifHandoffBackend.ts:530](packages/runtime-bridge/src/AifHandoffBackend.ts) `_rest`, is out of ownership and is reported above rather than left undiscovered. - **The vendored twins of every file touched**, plus the three that still carried #1618's defects — verified by reformatting the HEAD `src` of all ten vendored modules and diffing against the committed vendor, which isolates semantic drift from print-width drift. - **The getff payload manifest**: `cliEntry.ts` was absent from the first manifest rebuild because it was still untracked at build time, which would have shipped a vendored tree whose CLIs import a file the package does not deliver. Caught by re-running `build-getff-dist.sh --check` after the file was tracked; manifest now carries it. --- Prior-art: prior-art-evaluations.md#268 (node:util parseArgs, verdict ADOPT — built-in argv parsing replacing five hand-rolled copies; commander/yargs/minimist rejected in the same consult as runtime deps in a dependency-free vendored tree that runs from a bare consumer checkout with tsx as its only peer requirement). Prior-art: prior-art-evaluations.md#269 (es-main, verdict ADAPT — realpath-both-sides ESM main-module detection adopted as mechanism, not as a dependency; the repo already carried that implementation in cli/dispatch.ts and cli/claim.ts since #968).
artyhoo
added a commit
that referenced
this pull request
Sep 5, 2026
…1597 ledger A5-8) (#1627) ## Summary Ledger addendum **A5-8**, the tail of R-7 / S-4 (#1625): `AifHandoffBackend._rest` was the fourth hand-written copy of the aif-handoff request + BackendError mapping, and the one that survived the previous fold — so the contract still lived in two places, which is precisely the divergence R-7 documented (the copies had already grown different timeouts and a missing 429 branch). `_rest` is now a one-line delegation to `aifRequest` in [cli/aifHttp.ts](packages/runtime-bridge/src/cli/aifHttp.ts). ## Changes - **[AifHandoffBackend.ts:530](packages/runtime-bridge/src/AifHandoffBackend.ts)** — `_rest` body replaced by `aifRequest(method, this.baseUrl, path, body, { timeoutMs: REST_TIMEOUT_MS })`. The two properties this caller genuinely does not share with the CLIs are passed, not re-implemented: **DELETE** (the claim protocol issues one; the named aifHttp helpers do not cover it) and a **10 s timeout** instead of the CLI default 30 s, because a dispatch runs inside a PostToolUse hook that must not hold the author's editor — it falls back to ManualBackend. - **[cli/aifHttp.ts](packages/runtime-bridge/src/cli/aifHttp.ts)** — gains `aifRequest` (generic, method-taking, DELETE included) and an `AbortError` branch reporting **"timed out"** rather than `unreachable: The operation was aborted`. The latter is a real distinction for an operator reading a failure, and reachable by the CLIs only since #1625 gave them a timeout at all. - **Vendored twins** re-synced for BOTH changed files. Scope as assigned named only `vendor/src/AifHandoffBackend.ts`; `vendor/src/cli/aifHttp.ts` had to move with it, since a vendored backend importing an `aifRequest` its vendored aifHttp does not export is a copy that cannot run. **One operator-visible change, stated rather than buried:** backend REST failures now read `aif-handoff POST /tasks HTTP 500` where the private copy said `aif-handoff REST POST /tasks HTTP 500`. Nothing matches on that text (no `message.includes` or regex over it anywhere in `src/`), and adding a per-caller label option to the shared module to preserve the word is how a shared helper grows back into per-caller variants — so the shape is unified instead. ## Prior-art consult - [x] No capability commit here: no new file, no new dependency, no new capability — the commit carries the escape-hatch trailer with its rationale. - [x] The capability area (argv/HTTP plumbing for this tree) was already evaluated in this PR series: [prior-art-evaluations.md#268 / #269](docs/meta-factory/prior-art-evaluations.md), added in #1625. Nothing new surfaced. ## Test plan - [x] `npm test` in `packages/runtime-bridge` → **343 passed, 1 skipped** (was 341 on this base). - [x] `npx tsc --noEmit` clean; the **vendored tree typechecks standalone** with `vendor/` un-excluded (it sits outside the framework tsc scope by design). - [x] `npm run format:check` → clean. - [x] `bash scripts/build-getff-dist.sh --check` → in sync, 1061 files; the manifest delta vs `origin/staging` is exactly the 6 touched files, no additions or removals. ## Provenance n/a ## Review findings - **A5-8 — `_rest` is a fourth copy of the request + BackendError mapping.** Failure-scenario: the aif error contract lives in two functions in a tree that is vendored, so four shipped copies; adding a timeout or changing 429 handling in one leaves the others behind — which already happened before this series (the status probe aborted at 5 s while the CLI copy had no timeout at all, and had no 429 branch), and would happen again on the next change to either. Which test arm proves what (T3): - The **500 arm** in [aif-rest-dispatch.test.ts](packages/runtime-bridge/test/aif-rest-dispatch.test.ts) is RED on the pre-fold code: `origin/staging` `AifHandoffBackend.ts:576` emitted the `REST` prefix, so the asserted substring cannot match there. - The **429 arm is a preservation arm, not a RED** — that message was byte-identical in both copies (`origin/staging:570`), so it would have passed pre-fold too. It is there to pin that the fold did not move the quota classification. - The `aifRequest` arms in [aif-http.test.ts](packages/runtime-bridge/test/aif-http.test.ts) (DELETE, timed-out-vs-unreachable, caller-supplied timeout) were RED before the export existed. Not covered: nothing here ran against a live aif instance — all arms are against `fetch` doubles, same as the existing backend tests on this surface. ## Fidelity verdict FIDELITY: skipped — no kickoff: a single-finding follow-up dispatched from the #1597 review ledger addendum, not a stage of an umbrella ## Parked questions none ## §1.7 Self-discipline check (REQUIRED if PR touches discipline-bearing files) ### §1.7 Skipped: mechanical maintenance — this PR folds a duplicated private helper onto an existing shared module and touches no rule, principle, template or SSOT text
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Mechanical corrections to merged Stage 1 research-patches per Phase -1 Stage 1→2 cold-review (verdict REVISE, see
state.md §3.4). No verdict changes — SW-A REFERENCE (PR #268) and SW-B REJECT (PR #267) both unaffected.Findings closed by this patch
:39,:11751ce96a. Original record usedjq '.mergedAt'(camelCase, returns null); actual field is snake_casemerged_at. Verified viagh api repos/lee-to/aif-handoff/pulls/128 --jq '.merged_at':343e1ffa70) AND PR #128 (2026-05-26,51ce96a) are in the 30-day window. Both touchpackages/runtime/only:392ae7472a9→ae7072a9(aligned to line 119 body-text canonical version):210:17createTaskfunction…»)Caveats
curlagainst both URLs returns HTML shell with no extracted content. Pickedae7072a9because it appears in the patch body (line 119) where the worker documented the probe fresh from DeepWiki; the §9 «See also» list (line 392) is more prone to copy-paste typos. Falsifier: if a future session can verify against the actual DeepWiki search history and the canonical hash isae7472a9, this commit got it backwards.DN-1 / DN-2 still pending maintainer
This patch does NOT address:
SW-C dispatch remains BLOCKED on DN-1 + DN-2 answers + this PR landing on staging.
§1.7 Skipped: mechanical text correction to merged research-patches per Phase -1 Stage 1→2 cold-review findings B1/M1/m1/m2/m3. No new rule, principle, or discipline introduced; no SSOT verdict changed; no skill or template edited. Pure factual fixes verified against
gh apioutput (B1, M1) or inferred from internal consistency (m1). Therefore §1.7 forward/backward checks do not apply (matches meta-launch §4b «mechanical-maintenance escape hatch» pattern with rationale ≥60 chars).Test plan
gh api repos/lee-to/aif-handoff/pulls/128 --jq '.merged_at'returns2026-05-26T08:07:12Zgh api ...pulls/127,128