openkata-review-skill
Skill1 starsChanged 18 days agoMITView on GitHub
Lints, reviews, and optimizes a skill using the Tessl toolchain. Validates structure, evaluates quality, and applies improvements. Use when the user creates a new skill, updates an existing skill, asks to review or optimize a skill, or says "lint", "review", or "optimize" in the context of skills.
In this repository21 other files it gives its agents.
- AGENTS.mdAGENTS.md
- openkata-eval-runnerSkill
- openkata-optimize-skillsSkill
- openkata-review-ruleSkill
- openkata-rule-conventionsSkill
- openkata-ryu-releaseSkill
- openkata-skill-conventionsSkill
- clear-writingSkill
- commit-conventionsSkill
- create-adrSkill
- create-evalsSkill
- create-profileSkill
- create-ruleSkill
- create-skillSkill
- critical-thinkingSkill
- github-create-prSkill
- git-worktreeSkill
- grill-meSkill
- grill-with-docsSkill
- makefile-conventionsSkill
- spec-workflowSkill
The file
We split this skill into the rules it gives an agent
8 rules. About 0 are common, 5 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈5
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 12
Run the Tessl quality pipeline on a skill — lint, review, optimize.
- Distinctive: line 29
Read the diff and suggested changes.
- Distinctive: line 59
Generalize from feedback: Don't overfit fixes to specific cases.
- Very specific: line 62
Look for repeated patterns: If the agent keeps writing the same helper script or taking the same multi-step approach when using a skill, …
- Very specific: line 66
Explain the why: When adding instructions, explain reasoning instead of using heavy-handed ALWAYS/NEVER rules.
- Distinctive: line 69
Keep it lean: Remove instructions that aren't pulling their weight.
- Distinctive: line 85
Run tessl lint, review, and optimize on skills
- Distinctive: line 86
Run the skill design checklist
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Review Skill
Run the Tessl quality pipeline on a skill — lint, review, optimize. Works
on both new and updated skills in .agents/skills/ or skills/.
Workflow
- Identify the skill — Determine which skill to review from the user's request. Resolve the directory path. If ambiguous, list available skills and ask.
- Lint — Run
tessl skill lint <skill-directory>. Fix any structural issues before proceeding. - Review — Run
tessl skill review <skill-directory>. Read the full output — scores, suggestions, and warnings. This context is needed for changelogs and commits. - Optimize — Run
tessl skill review --optimize <skill-directory>. Read the diff and suggested changes. Apply improvements that make sense. If the interactive prompt blocks, apply the diff manually. - Re-lint — Run
tessl skill lint <skill-directory>again to confirm the optimized skill is still valid. - Report — Summarize what changed, the before/after review score, and any remaining warnings.
- Persist score — After review, update the skill's
.tessl-plugin/plugin.jsonwith the final score: read the JSON, set the"score"field to the numeric percentage, and write it back. This keeps scores queryable without re-running tessl. - Quality checklist — If
skills/create-skill/andskills/create-skill/references/skill-design-checklist.mdboth exist, run the checklist against the reviewed skill. Report findings. A missing## Boundariessection is a blocking failure — do not pass the review without it. If either path is missing, skip silently. - Feed learnings back — After every review, check whether the
insights could improve
create-skillor the skill design checklist. If a pattern keeps surfacing (e.g., a common mistake, a better convention), ask the user if they want to update the skill creation workflow so future skills benefit.
When Applying Improvements
- Generalize from feedback. Don't overfit fixes to specific cases. A skill will be used across many different prompts — ensure changes apply broadly, not just to the example that surfaced the issue.
- Look for repeated patterns. If the agent keeps writing the same
helper script or taking the same multi-step approach when using a
skill, bundle that into
scripts/so every future invocation benefits. - Explain the why. When adding instructions, explain reasoning
instead of using heavy-handed
ALWAYS/NEVERrules. Agents respond better to understanding why something matters. - Keep it lean. Remove instructions that aren't pulling their weight. If something isn't improving outputs, cut it.
Example Scenario
User: "Review the create-adr skill."
- Runs
tessl skill lint skills/create-adr— passes - Runs
tessl skill review skills/create-adr— 99% - Runs
tessl skill review --optimize— no changes needed - Runs checklist — all sections pass
- Reports: 99%, no blocking issues, one orphaned file warning
Boundaries
DOES:
- Run tessl lint, review, and optimize on skills
- Run the skill design checklist
- Suggest improvements and feed learnings back
Does NOT:
- Create new skills
- Publish or release skills
- Modify
create-skillwithout asking
Gotchas
- Tessl's optimize prompt is interactive — apply diffs manually when the prompt can't be answered.
- Markdown links in SKILL.md are validated by the linter as file paths relative to the skill directory. Use plain text references for files outside the skill folder.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

