Advisor ref-mode dispatch hangs silently from read-only surface #795

Closed
opened 2026-07-09 09:13:25 +00:00 by coilyco-ops · 2 comments
Member

Problem

From a read-only director surface, warded advisor coilyco-flight-deck/ward#792 ... resolved the issue ref and then did not return a broker-forwarded launch line or advisor result within multiple 30s intervals. The local surface wait had to be interrupted with Ctrl-C.

A second attempt with --thoroughness quick behaved the same way:

ward agent: resolved issue ref coilyco-flight-deck/ward#792 -> coilyco-flight-deck/ward#792

No comment appeared on ward#792 after the interrupted first attempt.

Expected behavior

A read-only director surface should be able to dispatch or run ref-mode advisor work without hanging opaquely. It should either:

  • forward to the host broker and print the dispatched log/path,
  • run the bounded advisor one-shot and post the answer, or
  • fail with a clear broker/launch error.

Do

  • Reproduce advisor ref-mode dispatch from a read-only surface through the broker.
  • Identify whether the hang is before broker forwarding, inside advisor container launch, inside one-shot capture, or waiting on a host-side process whose output is not surfaced.
  • Add a timeout or progress/error surface so the director is not left at a silent resolved-ref line.
  • Add regression coverage where practical.

Acceptance

  • warded advisor owner/repo#N ... from a read-only director surface either completes or fails loudly within the configured advisor timeout.
  • The surface prints enough information to find the host/container logs for a stuck advisor.
  • ward exec test passes.

Context

This surfaced while filing ward#792, the design question about replacing in-container review with brokered QA roles.

## Problem From a read-only director surface, `warded advisor coilyco-flight-deck/ward#792 ...` resolved the issue ref and then did not return a broker-forwarded launch line or advisor result within multiple 30s intervals. The local surface wait had to be interrupted with Ctrl-C. A second attempt with `--thoroughness quick` behaved the same way: ```text ward agent: resolved issue ref coilyco-flight-deck/ward#792 -> coilyco-flight-deck/ward#792 ``` No comment appeared on ward#792 after the interrupted first attempt. ## Expected behavior A read-only director surface should be able to dispatch or run ref-mode advisor work without hanging opaquely. It should either: * forward to the host broker and print the dispatched log/path, * run the bounded advisor one-shot and post the answer, or * fail with a clear broker/launch error. ## Do * Reproduce advisor ref-mode dispatch from a read-only surface through the broker. * Identify whether the hang is before broker forwarding, inside advisor container launch, inside one-shot capture, or waiting on a host-side process whose output is not surfaced. * Add a timeout or progress/error surface so the director is not left at a silent resolved-ref line. * Add regression coverage where practical. ## Acceptance * `warded advisor owner/repo#N ...` from a read-only director surface either completes or fails loudly within the configured advisor timeout. * The surface prints enough information to find the host/container logs for a stuck advisor. * `ward exec test` passes. ## Context This surfaced while filing ward#792, the design question about replacing in-container review with brokered QA roles.
Author
Member

🔒 Reserved by ward agent --harness codex — container engineer-codex-ward-795 on host kais-macbook-pro-2.local is carrying this issue (reserved 2026-07-09T09:14:37Z). Concurrent ward agent runs are blocked until it finishes or the reservation goes stale (1h TTL); --force overrides.

Do not comment on or edit this issue to steer the run while it is reserved. The engineer seeded the body once at launch and never re-reads it, so a comment or edit reaches only human readers, never the running engineer. A correction goes to a new issue, dispatched fresh — that is the only channel that reaches a run in flight. Where the forge supports it, ward locks this conversation to make that a road-block rather than a convention (ward#494).

run seed context — what this run is carrying (ward#609)
  • Resolved: coilyco-flight-deck/ward#795 · branch issue-795 · harness codex · workflow pr
  • Run: engineer-codex-ward-795 · ward v0.488.0 · dispatched 2026-07-09T09:14:37Z
  • Comment thread: 0 included in the pre-flight read, 0 stripped (ward's own automated comments).

Static container doctrine and seed boilerplate are identical every run and omitted here (they ride ward v0.488.0).

— Codex, via ward agent

<!-- ward-agent-reservation --> 🔒 Reserved by `ward agent --harness codex` — container `engineer-codex-ward-795` on host `kais-macbook-pro-2.local` is carrying this issue (reserved 2026-07-09T09:14:37Z). Concurrent `ward agent` runs are blocked until it finishes or the reservation goes stale (1h TTL); `--force` overrides. **Do not comment on or edit this issue to steer the run while it is reserved.** The engineer seeded the body once at launch and never re-reads it, so a comment or edit reaches only human readers, never the running engineer. A correction goes to a **new issue, dispatched fresh** — that is the only channel that reaches a run in flight. Where the forge supports it, ward locks this conversation to make that a road-block rather than a convention (ward#494). <details><summary>run seed context — what this run is carrying (ward#609)</summary> - **Resolved:** `coilyco-flight-deck/ward#795` · branch `issue-795` · harness `codex` · workflow `pr` - **Run:** `engineer-codex-ward-795` · ward `v0.488.0` · dispatched `2026-07-09T09:14:37Z` - **Comment thread:** 0 included in the pre-flight read, 0 stripped (ward's own automated comments). Static container doctrine and seed boilerplate are identical every run and omitted here (they ride ward v0.488.0). </details> <!-- ward-agent-signature --> — Codex, via `ward agent`
Author
Member

WARD-OUTCOME: done - advisor ref dispatch now acks immediately and no longer re-enters the broker on the host path.
workflow: pull-request; review summary: review gate skipped by --skip-review / --no-review-gate

The broker path now returns a log path right away, and the host-side launch runs with the broker-forwarding env stripped so it cannot recurse into itself. The only surprise was a repo-wide golangci-lint failure in unrelated workflow helpers; I cleaned that up too so the PR would go green. Confidence is high. The rough edge is that launch completion still lands in the background log, so a follow-up could add one more explicit surface breadcrumb if needed.

WARD-OUTCOME: done - advisor ref dispatch now acks immediately and no longer re-enters the broker on the host path. workflow: pull-request; review summary: review gate skipped by --skip-review / --no-review-gate The broker path now returns a log path right away, and the host-side launch runs with the broker-forwarding env stripped so it cannot recurse into itself. The only surprise was a repo-wide golangci-lint failure in unrelated workflow helpers; I cleaned that up too so the PR would go green. Confidence is high. The rough edge is that launch completion still lands in the background log, so a follow-up could add one more explicit surface breadcrumb if needed.
Sign in to join this conversation.
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
coilyco-flight-deck/ward#795
No description provided.