[Programcode]

This fixes the docs agents were reading wrong and adds mechanical checks so the same staleness gets caught next time.

Stale docs fixed

• docs/operations.md said staging E2E "has been skipped" since 2026-09-23. It actually runs after every main Quality push, and it failed today in authed-flows.spec.ts. The skipped entries come from PR-triggered Quality runs and are skipped by design. The paragraph now says how to check the live result instead of stating one.
• docs/operations.md also said PRs "may get preview deployments" without the catch: a Vercel preview uses the production Supabase project unless the in-app switcher picks staging (resolveSupabaseConfig in src/lib/supabase/env.ts). It now says so and points to the staging-pinned npm run dev:preview.
• plans/2026-10-01-voice-sparring-crossfire.md still said "Plan" although it shipped in feat(voice): spoken Sparring and Crossfire rounds #84. Four other plans had no status line at all. All six now have one.
• plans/README.md said shipped plan files "were removed" and then linked three it kept. That sentence now matches the list.

New checks

• npm run docs:check now requires a **Status:** line in the first 15 lines of every plans/*.md. It also rejects a plan linked under "## Shipped" whose status doesn't start with "Shipped". On main it reports exactly the five problems above. Two tests in check-docs.test.ts cover it.
• npm run backend:status now also reports:
  • the expected function secrets the project lacks, from supabase secrets list. It prints names only, and a test asserts the digests never reach the output;
  • the Auth site_url and redirect allowlist, read through the Management API with the CLI's existing access token. Without a token it says "not checked".

Agent pointers

• supabase/README.md gains an "Adding an Edge Function" checklist (config.toml, the db-gate and verify.sh readiness loops, the fake provider, --allow-env, EXPECTED_SECRETS). It also lists which secrets every deployed project needs and which ones fall back.
• AGENTS.md and plans/README.md now say to commit a plan before splitting it across agent threads.

Verification

• node scripts/check-docs.mjs on main flagged the five plan problems, and passes after the edits.
• npx vitest run src/test/docs/check-docs.test.ts src/test/scripts/backend-status.test.ts: 29 and 7 passed. The backend-status tests now isolate HOME and the access token, so they never call the real Management API.
• Ran backend:status read-only against both projects. Staging lacks STRIPE_SECRET_KEY, STRIPE_WEBHOOK_SECRET and ELEVENLABS_API_KEY. Production has every expected secret. Both allowlists printed.
• eslint and npm run typecheck pass.

Findings for Adam (no changes made)

• Staging's Auth site_url is http://localhost:8080.
• Production's redirect allowlist still includes two old Lovable domains (*.lovableproject.com, id-preview--*.lovable.app).
• Both projects hold secret names no code reads any more (ANTHROPIC_API_KEY, OPENAI_API_KEY, GOOGLE_*, AI_DEFAULT_*, plus JINA_API_KEY in production).
• The verify skill's readiness loop is missing voice and account, which db-gate.sh waits for.

Codex review (gpt-6.1-sol, high) raised two findings, both resolved:

• backend:status read only ~/.supabase/access-token. It now honors SUPABASE_HOME, and when no token file exists it says to set SUPABASE_ACCESS_TOKEN, which covers a token kept in the OS keyring.
• The docs referenced npm run ci:failures before chore(dev): per-worktree e2e port, ci:failures, db-gate --db-only, .nvmrc #97 added it. chore(dev): per-worktree e2e port, ci:failures, db-gate --db-only, .nvmrc #97 is merged and this branch includes it.

A review of the final head found nothing.

🤖 Generated with Claude Code

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
debate-navigator	Ready	Preview	Oct 2, 2026 7:40pm UTC

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[coderabbitai]

Warning

Review limit reached

Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available.
You're only billed for reviews past your plan's rate limits ($0.25/file).

Next included review available in 41 minutes.

Check out review usage here.

View limit details

Limit details: You’ve used the included review currently available.

Learn how review limits work.

Review configuration:

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Free

Run ID: f2aec3d1-0e7a-4135-9cb1-534719382d98

📥 Commits

Reviewing files that changed from the base of the PR and between a8a5167 and 473b1ff.

📒 Files selected for processing (13)

• AGENTS.md
• docs/operations.md
• plans/2026-09-20_131317-debate-workspace.md
• plans/2026-09-21-debate-workspace-no-bs.md
• plans/2026-09-25_074255-current-debate-foundation.md
• plans/2026-09-25_074255-debate-ai-quality-modules.md
• plans/2026-10-01-voice-sparring-crossfire.md
• plans/README.md
• scripts/backend-status.mjs
• scripts/check-docs.mjs
• src/test/docs/check-docs.test.ts
• src/test/scripts/backend-status.test.ts
• supabase/README.md

• Autopilot · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Autopilot is currently an internal CodeRabbit preview.

---

Note

🎁 Summarized by CodeRabbit Free

Your organization is on the Free plan. CodeRabbit will generate a high-level summary and a walkthrough for each pull request. For a comprehensive line-by-line review, please upgrade your subscription to CodeRabbit Essentials by visiting https://app.coderabbit.ai/settings/billing.

_{Comment @coderabbitai help to get the list of available commands.}

[Programcode]

live-e2e failed once on 473b1ff: e2e/live/research.spec.ts found a 2 px horizontal overflow on the Generate page after a real model run (npm run ci:failures -- 98). It depends on live model output, and this PR changes only docs, the docs checker and backend:status. Re-running the job. The overflow itself may be a real layout bug with some model outputs and is reported separately.