[developerz-ai]

What

A clean checkout has no packages/ai-claude-compat/dist (gitignored), yet that
package's only export is dist/index.js and packages/aitm imports it by package
name from dozens of modules. The node and integration CI jobs, plus the local
bin/setup and bin/check, now build the compat package before the authoritative
runners, so those runners resolve the runtime entry from a clean checkout
instead of depending on a pre-existing dist.

Acceptance

• .github/workflows/ci.yml: both the node and the integration job contain a step named build (compat runtime) running npm run build -w @developerz.ai/ai-claude-compat, placed before their test steps.
• bin/setup runs the compat build after bun install, and bin/check runs it before bun run test:node — grep finds npm run build -w @developerz.ai/ai-claude-compat in both.
• No bun test step exists anywhere in .github/workflows/ci.yml; the bun job still runs install/lint/madge/typecheck only, and the node and integration jobs remain the authoritative runners.
• With packages/ai-claude-compat/dist deleted, the repo gate (bin/check) completes and leaves packages/ai-claude-compat/dist/index.js on disk — the compat runtime entry exists with no manually generated artifact.
• bun run test:node and npm run test:integration -w @developerz.ai/aitm both pass with the compat entry produced only by the new step (no bun.lock, package.json or dependency change is part of this task).
• The command lines quoted in CLAUDE.md, .claude/skills/verify-change/SKILL.md, .claude/skills/integration-suite/SKILL.md and .dz/maintainer/reviewer.md match the new bin/setup, bin/check and CI step exactly.

Changes

• .github/workflows/ci.yml (edit) — Add a step named build (compat runtime) running npm run build -w @developerz.ai/ai-claude-compat to the node job (right after its install step, before its test step) and to the integration job (right after install, before integration tests): a clean checkout has no packages/ai-claude-compat/dist, yet that dist/index.js is the package's ONLY export and packages/aitm imports it by package name. Leave the bun job untouched and add no bun test step (Support the node:test built-in API oven-sh/bun#5090).
• bin/setup (edit) — Chain npm run build -w @developerz.ai/ai-claude-compat after bun install so a clean checkout ends setup with the compat runtime entry (packages/ai-claude-compat/dist/index.js) present — the authoritative node/integration runners then resolve @developerz.ai/ai-claude-compat without a manually generated artifact. Keep set -euo pipefail and the existing cd.
• bin/check (edit) — Insert npm run build -w @developerz.ai/ai-claude-compat && immediately before bun run test:node in the gate chain, so the authoritative node and integration runners always see the compat runtime entry even when bin/setup was never run. Every other step keeps its current order (typecheck → typecheck:tests → lint → madge → test:node → test:integration).
• CLAUDE.md (edit) — Fix the house tables this change falsifies: the bin/check row (add the compat build step AND the test:integration step it currently omits), the bin/setup row (bun install + compat build), and the ci.yml job-table rows for node and integration (both gain the compat build step). Quote the exact command strings from the shared spec; touch nothing else in the file.
• .claude/skills/verify-change/SKILL.md (edit) — Refresh the gate table this change falsifies: the install row (bin/setup now also builds the compat runtime entry) and the prove row (bin/check's command line now contains npm run build -w @developerz.ai/ai-claude-compat before bun run test:node). Quote the exact strings from the shared spec; leave the exit-code rules and the other rows alone.
• .claude/skills/integration-suite/SKILL.md (edit) — Make the run recipe true again: step 1 (install) must also build the compat runtime entry before the suite can resolve its imports, and step 6 (CI reference) must say the integration job in ci.yml now builds that entry before test:integration. Keep the rest of the recipe as is.
• .dz/maintainer/reviewer.md (edit) — Keep the maintainer brief's quoted gate true: the two places that spell out bin/check's command line (the Stack section and the 'Gates that matter' bullet) must include the compat runtime build step, in the same order as the script. Do not touch the other .dz briefs or .dz/onboarding/scorecard.md.

Verification

./bin/check passed (exit 0) in 3m 37s.

---

🤖 Opened by developerz.ai for task tsk_a603f70bf281e4f05b657ac0b05bbab7 on developerz-ai/ai-task-master.

Raised automatically by the raw_task_triage lane.

🤖 developerz.ai, an automated agent, not a human — this PR is one slice of a stacked plan.
Dispatched automatically by the raw_task_triage lane
Model MiniMax-M3 on router.ai.developerz.ai via native, funded by account_key
Ran on box hv-dedicated-dev-din
Cost unknown: no list price on record for MiniMax-M3 on router.ai.developerz.ai
Run timeline run_fbd03f9a4b4548a4b4a4e4ad5b3f359c
Part 1/1 of plan rtk_9d695c775b8c4c3990f08a88df1459b0

[developerz-ai]

Make compat runtime artifacts available before tests

• Picked up
• Surveying the repo — 2026-11-03 08:58 UTC
• Working — 2026-11-03 09:11 UTC
• Pull request opened — ci: build the compat runtime entry before node and integration tests #377 — 2026-11-03 09:44 UTC
• Done

Work landed on #377 — watching it toward merge.

46m · 126 tool calls · 21 messages

Sessions: coding run

---

🤖 developerz.ai, an automated agent, not a human — hv-dedicated-dev-din (developerz-ai[bot]) · model MiniMax-M3 on router.ai.developerz.ai

[developerz-ai]

✅ Reviewed — nothing blocking

0 actionable comment(s) · grounded on your code

_{⚠️ 1 config note(s): test files were not shown to the reviewer — this diff touched no path recognised as a test, so no assertion was available to read the change against}

_{⏱ 1m 01s wall clock · gpt-6.1-sol via router · 2 model call(s) · 634 output token(s)}

_{🤖 developerz.ai — automated review, running on your model and your box.}

[developerz-ai]

Review summary — 6 file(s), 0 finding(s).

• test files were not shown to the reviewer — this diff touched no path recognised as a test, so no assertion was available to read the change against

🤖 developerz.ai review — automated, running on your model and your box. What is this? · Reviewed by router/gpt-6.1-sol on box hv-dedicated-4 — 47.7k in / 976 out over 3 calls; cost unknown: no price on record for router/gpt-6.1-sol.

[developerz-ai]

Waiting on the repository owner: this pull request changes .dz/maintainer/reviewer.md, which the platform does not merge by itself.

The merge check refused the edit: .dz/maintainer/reviewer.md: no transform is declared for this gate path — only a workflow edit (T1, T2), an ADDED .github/workflows/ci.yml / .mcp.json / .dz/maintainer/reviewer.md (T3), a new .dz/maintainer/maintainer.yml (T4) or a MODIFIED .mcp.json (T5) has one for #377 — the PR is re-proved when its head changes, when the platform changes its gate-path transforms or templates, and on the hourly re-check; no .maintainer.yml key and no approval releases it — a workflow edit proves only as a tag pin, a permissions narrowing or a platform template, an added .mcp.json or reviewer prompt only as the platform template, a modified .mcp.json only as the platform adding its managed servers to the base file and touching nothing of the maintainer's, and a policy file only as a NEW .dz/maintainer/maintainer.yml on a repo with none that grants no more than the account's posture default

The task it answers planned this edit, so the goal cannot be met without it and it is not re-filed without them.

To take it, merge it on GitHub. To have the platform merge it, push a narrower edit: a new head is checked again and merged when the check passes. Nothing here is re-tried or closed automatically, and no email is sent; the pull request is listed with the stuck pull requests.

🤖 Posted by developerz.ai — an automated agent, not a human.

[developerz-ai]

Review summary — 6 file(s), 0 finding(s).

completed Adds the compat runtime build before local tests and updates the documented commands. No blocking defects found in the supplied diff; the reported CI jobs passed.

• test files were not shown to the reviewer — this diff touched no path recognised as a test, so no assertion was available to read the change against

🤖 developerz.ai review — automated, running on your model and your box. What is this? · Reviewed by router/gpt-6.1-sol on box hv-dedicated-dev-din — 46.1k in / 634 out over 2 calls; cost unknown: no price on record for router/gpt-6.1-sol.