[kyle-sexton]

Closes #806

Summary

When intake-triage (or any caller of the two-job pattern) falls back to the subscription token, the run's only lane record comes from the subscription job and does not say why it ran. This adds a machine-readable fallback=<reason> token so a credit stop on a role's API key shows in the lane record.

Fix

• report-lane-outcome: new optional fallback-reason input, credit-stop or no-key. recordTokens prints fallback=<reason> right after credential=, only for a value in FALLBACK_REASONS. The step script fails (core.setFailed, naming the allowed values, before any annotation or output) on any other non-empty value. The value reaches the script through env only.
• intake-triage.yml: the subscription job's "Report triage outcome" step passes fallback-reason: ${{ needs.claude.outputs.lane-class == 'credit-stop' && 'credit-stop' || needs.claude.outputs.credential-present == 'false' && 'no-key' || '' }}. That is a with: input on a uses: step, so the subscription job still writes no needs. value into a run: or script:.
• intake-triage-legs.test.cjs: the parity test strips exactly this one input from the subscription job (after asserting its exact text), and a new test evaluates it: credit-stop gives credit-stop, an empty key gives no-key, anything else gives empty.
• README: the run-record token list and the two-job pattern describe the input.

Out of scope, per the brief: an auth value; the claude-code-plugins follow-ups (pr-verify passing the input, the morning brief counting the token).

Verification

• node --test .github/scripts/*.test.cjs .github/actions/*/*.test.cjs: 632 pass, 0 fail.
• New tests were run red before the change: 3 in report-lane-outcome (record token order and vocabulary, notice and class line carry credential=... fallback=..., out-of-vocabulary value fails the step with no outputs), and 2 in the legs test (parity and expression evaluation) before the workflow edit.
• Existing record tests are unchanged and pass, so records without the input are byte-identical.
• actionlint, biome, typos and markdownlint-cli2 (README) are clean. No shell was touched.

Related

• intake-triage: give the model a read-only token; move label writes to a model-free job #804 / fix(intake-triage): run the model read-only and apply in a model-free job #821 (read-only model jobs) and fix(intake-triage): pass the model only comments from write-access authors #782 / fix(intake-triage): pass the model only comments from trusted authors #830 (trusted-comment filter): this is built on their structure, and nothing they added is relaxed.
• Downstream: feat(pr-pipeline): run pr-verify's model on the key first and the subscription token only on a credit stop claude-code-plugins#8577 (the workaround this replaces after a repin).

🤖 Generated with Claude Code

[claude]

Claude finished @kyle-sexton's task in 53s —— View job

---

Security review

• Gather context (PR diff, changed files)
• Review report-lane-outcome action.yml + classify.cjs for new fallback-reason input handling
• Review intake-triage.yml expression wiring
• Check instruction-surface / trust-boundary concerns
• Post findings

No security issues found.

Notes from the review (informational, not findings):

• The new fallback-reason input reaches the github-script step only through env.FALLBACK_REASON (action.yml#L163), never interpolated into a shell run: block, so there's no expression/command-injection surface there.
• intake-triage.yml's new line (intake-triage.yml#L1289) is a with: input on a uses: step, and the ternary only ever yields one of the two fixed literals 'credit-stop' / 'no-key' or '' regardless of what needs.claude.outputs.lane-class actually contains — so even an adversarial value there can't widen what reaches the composite action.
• action.yml validates fallback-reason against the fixed FALLBACK_REASONS vocabulary and calls core.setFailed (with no annotation/output written) before any reporting logic runs on an out-of-vocabulary value, so a caller wiring mistake fails closed rather than silently fabricating a token.
• classify.cjs's recordTokens independently gates on FALLBACK_REASONS.includes(...) before emitting the token, so even if the upstream check were ever bypassed, the printed record still can't carry an arbitrary string.
• No instruction-surface deletions (CLAUDE.md/AGENTS.md/rules files/hooks) in this diff — purely additive.

claude-lane-summary: lane=claude-security-review sha=5daaf70d25b429429c2dcb8653bf81fe1e8b1abf run=39502188889

[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 2026-10-20T21:40:56.270806Z	5daaf70	Draft marked ready

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.

[claude]

Claude finished @kyle-sexton's task in 4m 41s —— View job

---

Code review complete

• Gather context (diff, rule files, CLAUDE.md, README)
• Review report-lane-outcome action.yml + classify.cjs changes
• Review test coverage (classify.test.cjs, outcome-step.test.cjs)
• Review intake-triage.yml wiring + legs test
• Review README updates
• Check PR body contract (pr-body-contract.md rule)
• Post findings

Summary

No blocking or high-signal findings. This is a clean, well-tested change:

• classify.cjs's FALLBACK_REASONS closed vocabulary and recordTokens ordering (credential= then fallback=) match the PR description and the updated action.yml/README prose exactly.
• action.yml's validation (core.setFailed + return before touching execution-file or emitting any output/annotation) correctly rejects any value outside credit-stop/no-key, confirmed by reading the composite step script directly.
• The new fallback-reason: wiring in intake-triage.yml (line 1289) reads needs.claude.outputs.* inside a with: input on a uses: step, consistent with this file's existing script-injection avoidance for the two-job pattern; the expression is only reachable when the subscription job's own if gate (needs.claude.outputs.lane-class == 'credit-stop' || needs.claude.outputs.credential-present == 'false') already holds, so the '' branch of the ternary is dead by construction — not a bug, just confirms the two expressions are intentionally kept in sync.
• intake-triage-legs.test.cjs's new test reuses the existing actions-expression.cjs evaluator rather than inventing a new one, and its stripped-input parity test still asserts the exact wired text before deleting it.
• Test additions in classify.test.cjs and outcome-step.test.cjs run the actual shipped script/function (not a reimplementation) and cover the forged-value rejection path (case variants, leading space, embedded ::error::), the success-notice and failure-annotation record lines, and the empty/no-fallback case.

Not independently verified: I could not execute node --test in this sandbox (command execution required approval that wasn't available), so the PR's "632 pass, 0 fail" and the red-then-green TDD claims are author-stated and unverified by me — I verified correctness by reading the shipped script/function logic directly instead.

Unread: no .claude-lane/pr-context.md was present, so issue #806's acceptance criteria were not available to check against; nothing closing-issue-shaped was verified beyond the PR body's own description.

PR body contract: compliant — opens with Closes #806, includes ## Summary, ## Fix, ## Verification, ## Related.

claude-lane-summary: lane=claude-review sha=5daaf70d25b429429c2dcb8653bf81fe1e8b1abf run=39502188889

[github-actions]

claude-security-review has reviewed this pull request through 5daaf70; a later push is reviewed from there.

[github-actions]

claude-review has reviewed this pull request through 5daaf70; a later push is reviewed from there.