Apply mode has no liveness gate — unconfirmed pipeline roles generate wasted application materials

[ferrarilabs]

Code of Conduct

• I agree to follow the Code of Conduct

Existing issues

• I searched existing issues and this hasn't been reported yet

What happened?

Impact: High — applied significant time generating and attempting to submit applications for 8 roles, all of which were already closed/filled.

Root cause:
Batch scan adds roles to pipeline.md with **Verification:** unconfirmed (batch mode) because Playwright is unavailable in headless mode. When the user later enters apply mode, there is no liveness re-check step — it proceeds directly to generating materials against unverified URLs.

Proposed fix:

• apply mode should run check-liveness.mjs against each URL before generating materials
• Expired/closed roles should be auto-marked as Discarded and skipped before the user opens any tabs
• Or: add a post-scan liveness sweep for all unconfirmed pipeline entries before the next session

Workaround: Manually run node check-liveness.mjs against all unconfirmed pipeline entries before investing time in applications.

Steps to reproduce

Steps to Reproduce:

1. Run /career-ops scan (or node scan.mjs) — Playwright unavailable, falls back to WebSearch/WebFetch
2. New roles are added to pipeline.md with **Verification:** unconfirmed (batch mode)
3. User asks to apply to pending pipeline roles
4. apply mode generates tailored cover letters, Q&A answers, and application materials for all unconfirmed roles
5. User opens ATS URLs — roles are closed/filled/expired
6. All time and tokens spent on application materials are wasted

Expected behavior: Before generating any application materials, apply mode should call check-liveness.mjs on each URL. Roles that return expired/closed should be flagged immediately and skipped — user is told "X roles were closed, removed from pipeline" before any materials are generated.

Actual behavior: apply mode has no liveness gate. It generates full application packets for all roles regardless of verification status.

Environment:

• career-ops v1.8.1
• Remote execution environment (Claude Code web)
• Playwright not available (no browser in container)
• All 8 roles affected were marked unconfirmed (batch mode) in pipeline.md

Expected behavior

It should return only LIVE roles. I waste time and tokens in a daily basis with false results.

CLI tool

No response

OS and Node.js version

No response

Output of npm run doctor

[github-actions]

Thanks for opening your first issue, @ferrarilabs! We'll take a look soon.

In the meantime:

• Check SUPPORT.md for setup help
• Join our Discord for quick answers

[santifer]

Confirmed, @ferrarilabs — apply mode skips straight to generating materials with no liveness check, and wasted application packets are exactly the failure that hurts. Good news: check-liveness.mjs already exists, so this is a mode-file change to gate unconfirmed entries before generation. On the fix list (grouped with #835 as one liveness-gate class) — and if you want to PR it, scan.md step 7.5 shows the pattern.

[FReptar0]

Picked this up in #973, @ferrarilabs — hope you don't mind! It had been quiet for a couple of weeks and it slots right next to the liveness work I did in #487. Full credit for the diagnosis: your "8 dead tabs" repro is exactly the failure the batch sweep kills. Happy to hand it back or co-author if you were mid-flight on your own fix. 🙏