fix: create runs_dir only on write paths, not in RunManager.__init__

[ultimatile]

Summary

Metadata-lookup commands (status, list, job-output, wait) invoked against the wrong project root used to create an empty .crewster/runs tree there before printing Run not found, because RunManager.__init__ unconditionally ran mkdir(parents=True, exist_ok=True). Directory creation now happens only on write paths, and failed lookups leave the filesystem untouched. Closes #44

Changes

• src/crewster/run.py: removed the mkdir from RunManager.__init__; create_run and save_run_meta already create their run directory with parents=True, which covers runs_dir. list_runs treats a missing runs_dir as "no runs" — EAFP around iterdir, so a runs_dir that exists as a regular file still fails loudly.
• src/crewster/cli.py: new _print_run_not_found helper used by job-output and wait on a failed lookup. When no runs are recorded under the searched root, it names that directory and notes, conditionally, that a run submitted from a different project root must be looked up with the same --project-dir/--config as at submit time. The hint gates on list_runs (the manager's definition of a recorded run), so stray entries such as .DS_Store or a metadata-less directory cannot suppress it.

Impact

• submit is unaffected (its write path creates the directory).
• status / list / job-output / wait lose the mkdir side effect; load_run_meta still raises FileNotFoundError for a missing tree, which the id-lookup commands (status, job-output, wait) already catch, and crewster list in a fresh project still prints No runs found.

Test plan

Invariant: a failed metadata lookup never creates .crewster/runs.

• tests/test_run.py: __init__ creates nothing; list_runs / find_run_by_job_id / load_run_meta on a missing runs_dir return [] / None / raise without touching the filesystem; create_run and save_run_meta create the tree when missing.
• tests/test_cli.py: wait and status with an unknown id and list in a fresh project leave .crewster uncreated; the hint fires on an empty metadata tree, survives stray non-run entries, and is suppressed when runs are recorded.
• pytest: 379 passed; ruff check / ruff format / pyright clean.

Notes

• status still queries the scheduler when an id matches no local run metadata, without reporting the metadata miss; its Run not found branch is unreachable for a non-empty argument. Tracked in status: Run not found branch is unreachable after the job-id fallback #47.
• On a failed lookup the hint re-lists runs after find_run_by_job_id already did. The double scan on this error path is deliberate: it keeps the emptiness check inside list_runs instead of threading the run list from the lookup into the printing helper.

Summary by CodeRabbit

• Bug Fixes
  • Improved “Run not found” messages with guidance when no run history is available.
  • Read-only commands no longer create an empty .crewster directory.
  • list, status, and wait now handle projects without recorded runs more cleanly.
  • Preserved clear errors for missing runs and jobs while avoiding unnecessary filesystem changes.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: 83d9e0af-a422-418d-a264-95fd39fc7cf5

📥 Commits

Reviewing files that changed from the base of the PR and between 243eacb and 6cad027.

📒 Files selected for processing (4)

• src/crewster/cli.py
• src/crewster/run.py
• tests/test_cli.py
• tests/test_run.py

---

📝 Walkthrough

Walkthrough

Run metadata lookup paths no longer create .crewster/runs directories. Missing-run CLI responses now include a project-root hint when no metadata is recorded, with coverage for read and write lifecycle behavior.

Changes

Run lookup behavior

Layer / File(s)	Summary
Lazy runs directory lifecycle src/crewster/run.py, tests/test_run.py	RunManager defers directory creation, returns no runs when storage is absent, and preserves directory creation for write operations.
CLI missing-run feedback src/crewster/cli.py, tests/test_cli.py	job-output and wait use shared missing-run messaging, while CLI tests verify hints and filesystem behavior across lookup commands.

Estimated code review effort: 3 (Moderate) | ~20 minutes

Poem

I’m a bunny with a tidy run,
No empty folders in the sun.
Missing paths now softly say,
“Try the project used that day.”
Metadata blooms when writes begin!

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 60.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately summarizes the main change: deferring runs_dir creation to write paths instead of RunManager initialization.
Linked Issues check	✅ Passed	The PR addresses #44 by making runs_dir lazy, treating missing run directories as empty, and adding lookup hints/tests for metadata commands.
Out of Scope Changes check	✅ Passed	The changes stay focused on lazy runs_dir creation, lookup behavior, and related tests; no unrelated scope is evident.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix/44-lazy-runs-dir-mkdir

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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