Repository navigation
Open
Labels
effort/sSmall: expected under 2 hours of workSmall: expected under 2 hours of workkind/ciCI/CD pipelines, workflows, and build automationCI/CD pipelines, workflows, and build automationmain-brokenmain branch full-suite (main tier) is red; remediation chain owns thismain branch full-suite (main tier) is red; remediation chain owns thismeta/enrichedIntake review done: acceptance criteria, links, and normalized labels addedIntake review done: acceptance criteria, links, and normalized labels addedpriority/criticalDrop everything: main broken, security exposure, repo-wide outageDrop everything: main broken, security exposure, repo-wide outagesource/dispatchFiled by pm-dispatchFiled by pm-dispatchstatus/needs-triageNot yet triaged; needs a first lookNot yet triaged; needs a first look
Description
Main-tier workflow id failed on main.
- Run: https://github.com/developing-today/code/actions/runs/41720876149
- Head commit:
421250ee1b77685859af9cbc6233132f1884e4ff - Failing jobs: basic: nix flake check (biome-check),basic: nix flake check (cargo-fmt-check),basic: nix flake check (clippy-lint),basic: nix flake check (prettier-check),basic: nix flake check (rustfmt-check),basic: nix flake check (test-shard-0),basic: nix flake check (test-unit),basic: nix flake check (treefmt-check),basic: nix flake check (web-fmt-check),basic: nix flake check (web-lint),native arm64: world snapshot determinism,nix flake check
- Breakage-ID: 6e599ef68beb
Filed by the main-breakage chain (.github/workflows/main-broken.yml, #1782). Exactly one holder owns this breakage at a time; every transition is an agent-note:v1 comment on this issue.
Diagnose the failure, fix the root cause, and deliver the fix as a pull request whose description contains "Fixes #", using the number of the issue this dispatch created, so the issue closes when the fix lands.
Acceptance criteria
- Root cause of the main-tier failure diagnosed from the linked run's logs and fixed on main (or the breakage is shown to be superseded by a later green main run, with that run linked).
- The workflow named in this issue completes green on a subsequent main commit.
- The fix lands as a PR whose description contains "Fixes main-broken: developing-today/code id @421250e — basic: nix flake check (biome-check) #5300" so this issue closes on merge.
Metadata
Metadata
Assignees
Labels
effort/sSmall: expected under 2 hours of workSmall: expected under 2 hours of workkind/ciCI/CD pipelines, workflows, and build automationCI/CD pipelines, workflows, and build automationmain-brokenmain branch full-suite (main tier) is red; remediation chain owns thismain branch full-suite (main tier) is red; remediation chain owns thismeta/enrichedIntake review done: acceptance criteria, links, and normalized labels addedIntake review done: acceptance criteria, links, and normalized labels addedpriority/criticalDrop everything: main broken, security exposure, repo-wide outageDrop everything: main broken, security exposure, repo-wide outagesource/dispatchFiled by pm-dispatchFiled by pm-dispatchstatus/needs-triageNot yet triaged; needs a first lookNot yet triaged; needs a first look
Activity
developing-today commentedon Nov 4, 2026
Filed by pm-dispatch (GitHub Actions OIDC (repo developing-today/code, actor dezren39, workflow developing-today/code/.github/workflows/main-broken.yml@refs/heads/main)).
github-actions commentedon Nov 4, 2026
agent-note:v1 — holder: chain (filing), at 2026-11-04T05:50:10Z. Breakage-ID: 6e599ef68beb
github-actions commentedon Nov 4, 2026
agent-note:v1 — holder: chain (jules budget exhausted) at 2026-11-04T05:50:18Z: 3 open main-broken issues already carry the jules label. This issue carries queue/queued, so the fleet monitor enqueues the priority packet from the issue.
dezren39 commentedon Nov 4, 2026
🤖 Agent note · picked up
dezren39 commentedon Nov 4, 2026
🤖 Agent note · done
Findings — two independent failure classes, not one bug
Class 1 — real formatting drift at 421250e (code). The
rustfmt-checkjob log shows genuine rustfmt diff hunks against the head's own sources (e.g. a multi-lineChatError::Text(...)call and aplan_download_pathstest call that rustfmt rewrites). The same drift class explains the treefmt / cargo-fmt / prettier / biome / web-fmt failures in this run: unformatted content reached main.Class 2 — Magic Nix Cache throttling (infra, not code). The
test-unitandclippy-lintjob logs fail before any test/lint runs:error: unable to download 'http://127.0.0.1:37515/….narinfo': HTTP error 418withGitHub API error: Twirp error … code: ResourceExhausted, msg: "rate limit exceeded"andGitHub Actions Cache throttled Magic Nix Cache. Not trying to use it again on this run.Those jobs (and plausibly test-shard-0 / web-lint / world-snapshot in the same window) never evaluated the code at all.State since
Main has moved 189 commits past 421250e (current head d53b4b2). The range contains the drift response: two
style(treefmt): auto-format on mainsweeps (b65342b, cac5507),fix(id): format biome drift …(71dc2fd, PR #5123) and the biome editor-test format (6cb03fe, PR #5234). No further fix is writable from this triage: the drift class is already addressed downstream, and the cache class is transient infra.Confirmation condition (the one open thread)
Whether current main is fully green is not yet verified: every newer
idworkflow run on main (41767020520 … 41780242114) is stillqueued— the CI backlog has not produced a completed main run since the breakage. The first completed id-check run on main at/after d53b4b2 settles it: green closes this breakage out; a recurrence re-enters the main-broken chain under its own breakage-ID. Issue #5300 stays open for that confirmation rather than being closed on inference.dezren39 commentedon Nov 4, 2026
Intake review
Fleet monitor intake pass (protocol §5), 2026-11-04.
Related: the main-breakage chain (issue #1782); sibling main-broken issues are filed per breakage-id — this one owns its own breakage.
Notes: main-tier
idworkflow failure filed by the main-broken chain; priority/critical already applied. Acceptance criteria appended by this pass. Risks: flake-check failures at a single head commit can be superseded quickly — the worker should first check whether a later main run is already green before deep diagnosis. Suggested approach: check latest mainidrun; if still red, read the failing jobs' logs from the linked run and fix the root cause, not the symptom.