fix(cron): run scripts from job workdir

[ayahohner1]

What does this PR do?

Run cron script subprocesses from the job's configured workdir when one is present, while preserving the existing default of running scripts from HERMES_HOME/scripts when no workdir is configured.

Why this approach: the script path should still be resolved and contained under HERMES_HOME/scripts for safety, but the subprocess cwd should match the cron job's documented project workdir. Threading an explicit cwd into _run_job_script() fixes the bug without relying on process-global os.chdir().

Related Issue

Fixes #21721

Related existing PRs: #21397 and #42920. This PR is intentionally draft because it overlaps with existing contributor work; maintainers can close this if one of those should land instead.

Type of Change

• 🐛 Bug fix (non-breaking change that fixes an issue)
• ✨ New feature (non-breaking change that adds functionality)
• 🔒 Security fix
• 📝 Documentation update
• ✅ Tests (adding or improving test coverage)
• ♻️ Refactor (no behavior change)
• 🎯 New skill (bundled or hub)

Changes Made

• cron/scheduler.py
  • Add optional cwd parameter to _run_job_script().
  • Use that cwd for subprocess.run(...) when provided.
  • Resolve the job script cwd from job["workdir"].
  • Pass the job workdir to both no-agent scripts and agent pre-run scripts.
• tests/cron/test_cron_no_agent.py
  • Add regression coverage proving no-agent scripts can read project-relative files from workdir.
• tests/cron/test_scheduler.py
  • Add regression coverage proving agent pre-run scripts receive the configured workdir.

How to Test

1. Run the no-agent cron tests:

   python -m pytest tests/cron/test_cron_no_agent.py -q -o 'addopts='

   Expected: 19 passed

2. Run the targeted scheduler + no-agent cron tests:

   python -m pytest tests/cron/test_scheduler.py tests/cron/test_cron_no_agent.py -q -o 'addopts='

   Expected: 227 passed

3. Run lint on changed files:

   ruff check cron/scheduler.py tests/cron/test_cron_no_agent.py tests/cron/test_scheduler.py

   Expected: No issues found

Checklist

Code

• I've read the Contributing Guide
• My commit messages follow Conventional Commits (fix(scope):, feat(scope):, etc.)
• I searched for existing PRs to make sure this isn't a duplicate
• My PR contains only changes related to this fix/feature (no unrelated commits)
• I've run pytest tests/ -q and all tests pass
• I've added tests for my changes (required for bug fixes, strongly encouraged for features)
• I've tested on my platform: Ubuntu 24.04 / Linux 6.8.0-101-generic

Documentation & Housekeeping

• I've updated relevant documentation (README, docs/, docstrings) — or N/A
• I've updated cli-config.yaml.example if I added/changed config keys — or N/A
• I've updated CONTRIBUTING.md or AGENTS.md if I changed architecture or workflows — or N/A
• I've considered cross-platform impact (Windows, macOS) per the compatibility guide — or N/A
• I've updated tool descriptions/schemas if I changed tool behavior — or N/A

For New Skills

N/A

Screenshots / Logs

Pytest: 19 passed

Pytest: 227 passed
Ruff: No issues found

git diff --check: clean

[alt-glitch]

This was generated by AI during triage.

Duplicate of #21397 (earliest open PR) — both thread the configured cron job workdir into _run_job_script()'s subprocess cwd, overriding the hardcoded cwd=path.parent. Same function, same mechanism. Related to the no_agent-specific siblings #42274 / #56005 and the origin bug #21721. You flagged this yourself as a draft to avoid competing; keeping #21397 as canonical for the general fix.

[teknium1]

Thanks for the focused cron fix. The underlying issue is still present: current main hardcodes the script subprocess cwd at cron/scheduler.py:2106, and the agent pre-run path reaches it through cron/scheduler.py:2758.

Problems

• Current main added _run_job_script_with_claim_heartbeat() in cd537187611769ebb6a1aa9460265e9ef5694606; both no-agent and pre-run scripts now use it (cron/scheduler.py:2611, 2758). The salvage must thread the cwd through that wrapper rather than replacing those calls with direct _run_job_script() calls, or one-shot scripts lose their claim-refresh protection.
• Runtime workdir validation currently happens at cron/scheduler.py:2868-2876, after the pre-run script. Resolve and validate the cwd before both script paths so a removed workdir retains the existing script-directory fallback.
• The agent regression test mocks _run_job_script, but production now calls the heartbeat wrapper. Update it to assert the current integration path and retain real-subprocess coverage for both modes.

Suggested changes

• Pass a validated optional cwd through _run_job_script_with_claim_heartbeat() to _run_job_script() and remove the no-agent os.chdir() path.
• Add a claimed one-shot regression test to ensure the heartbeat wrapper remains active.

Automated hermes-sweeper review.

[teknium1]

Current main now routes no-agent scripts through _run_job_script_with_claim_heartbeat() to keep one-shot claims alive (cd5371876). During salvage, preserve that wrapper and thread cwd through it; replacing the call with direct _run_job_script() would drop the duplicate-dispatch protection.