Agent Skills: Git Worktree Status

Git Worktree Status

Report the background verification jobs launched by /git-worktree, without blocking. A Python script detects the worktree, reads .worktree-logs/*.log, and classifies each job using language-neutral markers (works for pytest, vitest, tsc, basedpyright, cargo, go). It is read-only — it never starts, stops, or re-runs a check.

Steps

Run from inside any worktree directory:

uv run "${CLAUDE_SKILL_DIR}/scripts/git_worktree_status.py"

scripts/git_worktree_status.py takes no arguments. Run it from inside a worktree; everything it needs comes from git and the log files. It:

• confirms the working directory is a worktree by inspecting git's common directory, and exits non-zero without a report if run from the main repo, outside git, or where git is unavailable;
• reads three known logs at the worktree top level — .worktree-logs/typecheck.log, .worktree-logs/tests.log, and .worktree-logs/build.log — each independently, with a missing log treated as NOT_RUN rather than an error;
• classifies each log from its captured text (not exit codes), reporting PASS, FAIL, RUNNING, or NOT_RUN, and returns immediately so a job still writing its log simply shows as RUNNING.

The logs themselves are produced by the background jobs that git-worktree launches during worktree setup; this skill only reads them. See references/status-semantics.md for the exact markers behind each status and how to interpret failures.

Status meanings

| Status | Meaning | | --------- | ------------------------------------------------------------ | | PASS | Log shows zero failures/errors or an explicit success marker | | FAIL | Log shows failures, errors, a traceback, or a panic | | RUNNING | Log exists but has no terminal marker yet | | NOT_RUN | No log file (the job was never started or logs were cleared) |

Numeric counts win first: a log reporting 0 failed is a PASS even though it contains the word "failed". For the precise per-status markers and the failure-mode rundown (not a worktree, missing log, truncated log, where to look on FAIL), see references/status-semantics.md.

Reference Files

• references/status-semantics.md — precise PASS/FAIL/RUNNING/NOT_RUN definitions, the language-neutral markers, which logs are read, why the report is non-blocking, and failure modes.

Re-running a check is not this skill's job — the commands that write these logs are owned by the git-worktree setup guides so the launch detail lives in one place. To re-launch a check, clear the relevant log and start it from the guide for your stack:

• Python: ../git-worktree/references/setup-python.md
• Node.js: ../git-worktree/references/setup-node.md
• Rust: ../git-worktree/references/setup-rust.md
• Go: ../git-worktree/references/setup-go.md

Related skills

This skill is the status reader in the worktree lifecycle suite. /git-worktree creates a worktree and launches the background checks; the rest of the suite manages those worktrees:

• /git-worktree — create a worktree and launch the verification jobs
• /git-worktree-status — read those jobs' status (this skill)
• /git-worktree-clean — clean up worktree artifacts
• /git-worktree-remove — remove a worktree
• /worktree-doctor — diagnose worktree health

Usage

/git-worktree-status

No arguments. Run from inside a worktree directory.