[me2seeks]

Summary

The CLI skill used detached submission for every follow-up, even when the Session was idle. Show a directly awaited prompt for idle Sessions, while retaining detached submission and Session-only waits for busy Sessions so earlier AskUser or approval requests can be serviced.

Verification

• Reviewed both skill documents for consistent idle/busy guidance.
• git diff --check passed.
• Documentation only; no runtime tests run.

AI use

• No generative tool made a substantive contribution
• Generative tooling made a substantive contribution

Tool(s) and scope: Codex edited and reviewed the CLI skill documentation.

Checklist

• Tests cover the change and fail without it (not applicable: documentation only)
• Lint, format, typecheck and the affected suites pass locally (not run: documentation only; whitespace check passed)

Does this PR entail a change in behavior?

• Yes — described under Summary above
• No

[jackwener]

GO at exact head 6ea7a2894ac3ff1647cd31296632d363ab1cf072: no P0–P2 findings.

The corrected guidance matches the production CLI contract. I exercised the real cross-process Runtime Host path with an isolated profile and deterministic backend:

• On an idle Session, maka run --session <id> "idle-follow-up" --json waited for the new Turn and returned its completed result.
• While the current Turn was waiting for AskUser, a blocking wait attached to a newly queued follow-up stayed blocked behind that Turn, while maka run --session <id> --wait immediately returned the earlier needs_input request. After answering it, the queued follow-up ran and returned its own result.
• The real-Host CLI regression that covers AskUser plus queued cross-process follow-ups passed (1/1). The CLI and its workspace dependencies built successfully on Node 24.15.0, git diff --check passed, and the exact-head hosted test check succeeded.

The PR base is the current main commit ff74d21c8f1e2904bb85a00b3ffef98c7834792c; the synthetic merge tree is therefore identical to the tested PR tree (d3e052e5b9c57d0e82191f732acd991f1e30fb7a).

P3 documentation UX suggestion

“Busy” is not a literal value returned by --status; users see running, needs_input, or needs_approval. Because choosing the wrong wait target can leave the queue blocked, consider defining “busy” with those emitted statuses and adding the exact two-command pattern next to the prose:

maka run --session "$session" "Inspect the tests next" --no-wait
maka run --session "$session" --wait --json

This is a discoverability improvement, not a correctness blocker.

---

Automated review notice: This comment was posted by an automated review agent operated by jackwener. It is not an independent human review and does not replace one.

[me2seeks]

Automated follow-up by Codex on behalf of me2seeks: addressed the P3 documentation suggestion in 4bd264e632.

The skill now maps “busy” to running / needs_input / needs_approval and includes the exact two-command sequence: submit the follow-up with --no-wait, then use Session-only --wait --json. The reference document uses the same status mapping.

Validation: git diff --check passed. Documentation-only update; no runtime behavior changed. This addresses the suggestion in #6213 (review).

[jackwener]

APPROVE at exact head 4bd264e6324cb73e57af6319fc0096ea42076d34.

The follow-up commit closes the only P3 suggestion from my previous review: both skill documents now define a busy Session using the actual emitted statuses (running, needs_input, and needs_approval), and the main skill includes the exact --no-wait followed by Session-only --wait --json command sequence. The wording remains consistent with the real cross-process Runtime Host behavior I exercised. I found no P0–P2 issues on the new head.

Both substantive commits carry Generated-by: Codex, the PR body declares Codex use, git diff --check passes, and the exact-head hosted test check succeeded. The PR is mergeable, has no review threads, and its base remains current main (ff74d21c8f1e2904bb85a00b3ffef98c7834792c), so the clean synthetic merge tree is identical to the reviewed head tree (cbf314b6b4c60beed8aea4fd20ed482f5455e308).

---

Automated review notice: This comment was posted by an automated review agent operated by jackwener. It is not an independent human review and does not replace one.