jujutsu
**REQUIRED** - Always activate FIRST on any git/VCS operations (commit, status, branch, push, etc.), especially when HEAD is detached. If `.jj/` exists -> this is a Jujutsu (jj) repo - raw git commands can corrupt data. Essential git safety instructions inside. DO NOT IGNORE.
80
74%
Does it follow best practices?
Impact
88%
1.27xAverage score across 3 eval scenarios
Advisory
Suggest reviewing before use
Optimize this skill with Tessl
npx tessl skill review --optimize ./jujutsu/SKILL.mdQuality
Discovery
72%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 description excels at trigger terms and distinctiveness — it's very clear about WHEN to activate and targets a specific niche (Jujutsu repos where raw git can corrupt data). However, it's weak on describing WHAT the skill actually does beyond 'safety instructions,' and the urgent, all-caps tone ('DO NOT IGNORE', 'REQUIRED') reads more like a system prompt directive than a skill description. It would benefit from briefly stating the concrete actions or guidance it provides.
Suggestions
Add specific capability descriptions: e.g., 'Translates git commands to jj equivalents, prevents raw git operations in Jujutsu repos, provides safe alternatives for commit/branch/push workflows.'
Tone down the imperative/alarm language ('DO NOT IGNORE', 'REQUIRED') and instead describe concrete actions in third person, e.g., 'Detects Jujutsu repositories and provides safe VCS command translations to prevent data corruption.'
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | It names the domain (git/VCS operations) and mentions specific operations (commit, status, branch, push), but doesn't describe concrete actions the skill performs beyond 'safety instructions.' It's more about when to activate than what it does. | 2 / 3 |
Completeness | The 'when' is very clearly stated (git/VCS operations, detached HEAD, .jj/ exists), but the 'what' is weak — it only says 'Essential git safety instructions inside' without explaining what those instructions actually do or what actions the skill performs. The explicit trigger guidance is strong, but the capability description is vague. | 2 / 3 |
Trigger Term Quality | Excellent coverage of natural trigger terms: 'git', 'VCS operations', 'commit', 'status', 'branch', 'push', 'HEAD', 'detached', '.jj/', 'Jujutsu', 'jj'. These are terms users would naturally use when working with version control. | 3 / 3 |
Distinctiveness Conflict Risk | Very distinct niche: Jujutsu (jj) repo safety when git commands are used. The .jj/ directory check and the specific focus on preventing data corruption in jj repos makes this highly distinguishable from general git skills. | 3 / 3 |
Total | 10 / 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 that provides comprehensive coverage of Jujutsu VCS operations with excellent agent-specific guidance. Its main strength is the concrete, executable commands paired with clear warnings about interactive operations that fail in agent environments. The primary weakness is that it's somewhat long for a single file, with some sections that could be more concise or split into supporting files for better progressive disclosure.
Suggestions
Consider splitting detailed sections (Git Integration, Refining Commits) into separate reference files and linking from the main SKILL.md to reduce token footprint.
Trim the Core Concepts section by removing explanations Claude can infer (e.g., 'There is no staging area', 'Changes are automatically snapshotted') and focus only on the non-obvious behavioral differences from git.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is mostly efficient but includes some unnecessary explanations (e.g., 'In jj, your working directory is always a commit' and 'There is no staging area' are concepts Claude can infer). The 'Core Concepts' section could be tighter, and some sections like 'Preserving Commit Quality' repeat guidance already implied. However, most content earns its place with concrete commands and agent-specific warnings. | 2 / 3 |
Actionability | Nearly every section provides concrete, executable commands with specific flags and arguments. The skill includes copy-paste ready examples for all major operations, agent-specific warnings about interactive commands that will hang, and clear alternatives (e.g., use `jj restore` instead of `jj split`). The quick reference table is immediately usable. | 3 / 3 |
Workflow Clarity | The essential workflow is clearly sequenced (describe first, then code, then check status). Validation checkpoints are explicit ('Verify operations with jj st after mutations'), there are feedback loops for conflict resolution and commit quality review, and the push workflow includes a pre-push checklist. Agent-specific hazards (interactive commands, editor prompts) are clearly flagged with alternatives. | 3 / 3 |
Progressive Disclosure | The content is well-organized with clear section headers and a logical progression from concepts to workflows to advanced operations. However, at ~250 lines with no bundle files or external references, some sections (like the full Git integration details or the detailed rebasing examples) could be split into separate reference files to keep the main skill leaner. The quick reference table partially compensates by providing a scannable summary. | 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 |
|---|---|---|
allowed_tools_field | 'allowed-tools' contains unusual tool name(s) | Warning |
Total | 10 / 11 Passed | |
- Repository
- danverbraganza/jujutsu-skill
- Commit
72207e4
- Reviewed
Table of Contents
Is this your skill?
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.