Ship the full current in-progress work as a PR in one pass. If already running inside a workspace-linked `git worktree` (for example under `.stave/workspaces/*`), reuse that same worktree in place. Otherwise move the dirty state into a dedicated temporary worktree, commit with a Conventional Commit message, push, open a GitHub pull request, and clean up the temporary worktree unless the user asks to keep it. Use for prompts like "worktree 만들어서 PR", "현재 작업 PR로 올려줘", or "spin this dirty tree into a PR branch".
92
88%
Does it follow best practices?
Impact
100%
2.17xAverage score across 3 eval scenarios
Passed
No known issues
Quality
Discovery
100%Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.
This is a strong, well-crafted description that clearly articulates a specific workflow (shipping dirty state as a PR via git worktrees), lists concrete actions in sequence, and provides explicit trigger phrases including multilingual examples. The conditional logic (reuse existing worktree vs. create temporary one) adds helpful specificity for skill selection.
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | Lists multiple specific concrete actions: move dirty state into a worktree, commit with Conventional Commit message, push, open a GitHub pull request, clean up the temporary worktree. Also describes conditional behavior (reuse existing worktree vs. create temporary one). | 3 / 3 |
Completeness | Clearly answers both 'what' (ship in-progress work as a PR via worktree, commit, push, open PR, clean up) and 'when' (explicit 'Use for prompts like...' clause with example trigger phrases). | 3 / 3 |
Trigger Term Quality | Includes strong natural trigger terms like 'PR', 'worktree', 'dirty tree', 'PR branch', 'git worktree', 'commit', 'push', and even Korean-language trigger phrases. These cover multiple natural ways a user might phrase the request. | 3 / 3 |
Distinctiveness Conflict Risk | Highly distinctive niche combining git worktrees with PR creation from dirty/in-progress state. The specific workflow (worktree-based PR shipping) is unlikely to conflict with generic git or PR skills. | 3 / 3 |
Total | 12 / 12 Passed |
Implementation
77%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a strong, highly actionable skill with excellent workflow clarity and concrete git commands throughout. Its main weakness is moderate verbosity—the guardrails section largely restates constraints already embedded in the workflow steps, and some substeps could be tightened. The progressive disclosure is adequate for a standalone file but the length suggests some content could be better organized or split.
Suggestions
Deduplicate the Guardrails section by removing items already explicitly covered in the workflow steps (e.g., --force-with-lease, worktree removal cwd constraint), keeping only constraints not obvious from the workflow.
Remove the introductory paragraph which restates the skill description and adds no new information for Claude.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is fairly detailed and well-structured, but it's verbose in places—some guardrails repeat what's already stated in the workflow steps (e.g., force-with-lease, worktree removal from inside). The two-sentence intro paragraph restates the description. However, most content is genuinely instructive and not explaining things Claude already knows. | 2 / 3 |
Actionability | The skill provides specific, executable git commands throughout (e.g., `git stash push --include-untracked -m "<message>"`, `git worktree add -b <branch> ../.worktrees/<repo>/<branch> HEAD`, `gh pr create --base <base> --head <branch> --title <title> --body <body>`). Decision logic is concrete with clear mode selection criteria and specific examples of PR title format. | 3 / 3 |
Workflow Clarity | The 9-step workflow is clearly sequenced with explicit validation checkpoints (step 4 validates changes, step 5 confirms clean status after commit, step 8 has retry logic for cleanup failures). It includes feedback loops for conflicts (step 3: stop and surface), coherence checks (step 5: stop and ask), and error recovery throughout. The two execution modes are clearly distinguished at each step. | 3 / 3 |
Progressive Disclosure | The content is well-organized with clear sections (Workflow, Guardrails) and numbered steps, but it's a monolithic document with no references to external files. Given the length (~120+ lines of detailed workflow), some content like the guardrails or the detailed mode-specific substeps could benefit from being split out or collapsed. However, with no bundle files provided, this is a reasonable single-file approach. | 2 / 3 |
Total | 10 / 12 Passed |
Validation
90%Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.
Validation — 10 / 11 Passed
Validation for skill structure
| Criteria | Description | Result |
|---|---|---|
frontmatter_unknown_keys | Unknown frontmatter key(s) found; consider removing or moving to metadata | Warning |
Total | 10 / 11 Passed | |
e275ad5
Table of Contents
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.