[kixelated]

Plans follow-ups from moq-dev/moq#7161, moq-dev/moq#7076 and moq-dev/moq#7166.

Quests

• quest/m1/clock-test-process.md [XS]: cargo test -p moq-clock fails on main every time. paused_runtimes_own_their_anchors assumes it is the first test in its process to read the clock, and passes only under nextest, which runs one process per test. Reproduced 2026-10-29.
• quest/m1/kio-workspace-tests.md [S]: two kio log eviction tests fail only in cargo test --workspace. They pass with -p kio, with and without _sim, so another unified dependency feature is the trigger. ci: unify dependency features with a cargo-hakari workspace-hack #6985's workspace-hack excludes kio, so it won't fix this.
• quest/m1/js-cache-age-test.md [XS]: broadcast.test.ts "a stalled consumer does not pin evicted groups" moves the system time, which the cache never reads, so the test can't fail.
• quest/m1/broadcast-clock.md [S]: every group a moq-mux broadcast writes is stamped from the broadcast's own clock. Today catalog, MSF and HLS record groups stamp with the ambient Timestamp::now(), so a call from another runtime can move a track's timestamps backwards.
• Moved m2 → m1: crates-trusted-publisher.md and release-rs-oidc.md. CARGO_REGISTRY_TOKEN is the last long-lived registry token.
• quest/m0/ci/land.md: adds an open question. May main keep deploying to the publish Environment while its ruleset lets role 2 (and, after land, moq-bot) bypass review?

Decisions

• Test fixes:
  • ✅ Three m1 quests
  • One m1 quest
  • m0 test-flakes-2 children
• Catalog timestamps going backwards:
  • Stamp with max(now, last)
  • ✅ The broadcast owns one clock (the maintainer: "We should be using one clock for the entire broadcast. That seems like the real bug")
  • Only fix moq-ffi's mixed clock contexts
• crates.io OIDC:
  • ✅ Move both quests to m1
  • Move both to m0
  • Leave them in m2
• publish Environment on main:
  • ✅ An open question in land.md
  • A separate m0 decision quest

🤖 Generated with Claude Code

(Written by Claude Opus 5.5)

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-29T04:16:25.260035Z	0bee112	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[kixelated]

Merging: Codex reviewed head 0bee112 with no findings (👍). quest check --base origin/main and sh/markdown.sh check pass locally. No changes made during merge; the plans are as settled through /quest-plan.

(Written by Claude Opus 5.5)

[kixelated]

Automated review by review (OpenAI)

Reviewed commit: 0bee112

No actionable bugs found in these planning changes. Separate test-isolation and clock-ownership quests are appropriately scoped: the clock plan includes the ambient reads inside Clock itself and explicitly requires reconciling timestamp semantics, rather than merely clamping regressions. The test plans require preserving the original guarantees and demonstrating meaningful failures. Moving trusted-publisher setup alongside the OIDC implementation keeps its prerequisite visible, while the publishing-environment decision remains explicitly with the maintainer.

Verification: static comparison with the referenced clock and stalled-consumer tests, clock/stamping paths, release workflow, and moved quest links, including an independent clock-plan review. I did not reproduce the reported failures, run tests, or inspect/change external registry or repository settings.