create-a-skill
Skill1 starsChanged 5 months agoMITView on GitHub
Interview, stress-test, and author new skills for the skills-i-use repo — or improve existing ones. Asks hard questions, runs reviewer subagents, writes design briefs, generates SKILL.md and supporting files, validates conventions, and registers in README and OUTPUT.md. Triggers on "create a skill", "new skill", "add a skill", "improve this skill", "author a skill", "skill for X workflow", "build a skill".
10 other projects use a version of this.
Adopted by
Most recently taken up by
- nguyenthanhtat/screen1-claudetheir own version, first seen Nov 1, 2026
- Future-CX/AI-Architecture-Toolkittheir own version, first seen Nov 1, 2026
- Pyroboomka/skillstheir own version, first seen Oct 25, 2026
- ubik-dsh/deepseek-harness-skillstheir own version, first seen Oct 11, 2026
- L-X-T/ng-agentic-skillstheir own version, first seen Oct 4, 2026
- ubik-dsh/deepseek-modstheir own version, first seen Sep 26, 2026
Other versions10 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 10
- L-X-T/ng-agentic-skills★ 13same name
- Future-CX/AI-Architecture-Toolkit★ 8same name
- navikt/hovmester★ 2same name
- nguyenthanhtat/screen1-claude★ 2same name
- shiggydoodah/skills★ 2same name
- supercharge-growth/ai-design-skills★ 2same name
- Pyroboomka/skills★ 1same name
- VodouAI/OS★ 1same name
- ubik-dsh/deepseek-harness-skills★ 0same name
- ubik-dsh/deepseek-mods★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository51 other files it gives its agents.
- architect-enterprise-aiSkill
- author-agent-skillsSkill
- build-agent-harnessesSkill
- build-aiewf-2024-ragSkill
- build-generative-mediaSkill
- build-multimodal-productsSkill
- build-rag-search-stacksSkill
- build-robotics-aiSkill
- build-with-deepmindSkill
- build-with-microsoft-aiSkill
- build-with-openaiSkill
- deploy-public-sector-aiSkill
- design-agent-skillsSkill
- evaluate-with-braintrustSkill
- lead-ai-engineering-teamsSkill
- lead-ai-transformationSkill
- learn-summit-2023-remoteSkill
- learn-summit-2023-talksSkill
- navigate-aie-europe-2026Skill
- navigate-aiewf-2024-keynotesSkill
- observe-ai-productionSkill
- operate-agent-memorySkill
- operate-legal-ai-agentsSkill
- operate-openclaw-agentsSkill
- optimize-llm-inferenceSkill
- plan-sovereign-aiSkill
- price-ai-productsSkill
- run-aiewf-2024-evalsSkill
- run-llm-evalsSkill
- run-summit-2023-workshopsSkill
- secure-ai-agentsSkill
- ship-ai-as-pmSkill
- ship-aiewf-2024-agentsSkill
- train-llms-locallySkill
- architecture-reviewSkill
- chat-failure-auditSkill
- cross-model-reviewSkill
- cto-weekly-reviewSkill
- friction-auditSkill
- ig-carouselSkill
- pdf-reportSkill
- raise-prSkill
- reel-editSkill
- research-reportSkill
- ship-checkSkill
- spec-grounded-designSkill
- state-lifetime-decisionSkill
- ste100Skill
- scroll-engineeringSkill
- ui-uxSkill
- verify-like-a-userSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
13 rules. 0 are common, 11 are distinctive, and 2 are very specific to this project.
0
Common
Most of the 8 edited versions we read keep these.
11
Distinctive
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 25
Scripts: read scripts/validate-skill.sh before running.
- Distinctive: line 45
Read questions.md. Ask required + hard questions in one message.
- Distinctive: line 47
Do not proceed until required items are answered.
- Distinctive: line 50
Scan existing skills: before intake completes:
- Distinctive: line 59
If user provided verbatim wording for the skill, preserve it exactly in output files.
- Distinctive: line 85
Read reviewers.md. Run Skeptic, Maintainer, and Practitioner in parallel with the skill brief.
- Distinctive: line 90
If REJECT — explain why and suggest alternative (script, rule, extend X).
- Distinctive: line 92
If EXTEND — read existing skill folder fully before writing.
- Very specific: line 110
Create skills/[skill-name]/ using templates/ as starters.
- Distinctive: line 123
Read entire existing folder
- Distinctive: line 148
Update registry files (skip if improving and row already exists):
- Distinctive: line 170
Ask: "Commit to repo?" — only on explicit yes
- Distinctive: line 177
Ask separately before push.
“Common” is measured: kept by at least 6 in 10 of the 8 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Create-a-Skill — Author & Improve Repo Skills
End-to-end skill authoring for skills-i-use: collect → challenge → review → write → validate → register in repo.
Works with any coding agent that can run subagents. Follows CONTRIBUTING.md.
Supporting files:
- questions.md — intake and hard questions
- reviewers.md — Skeptic / Maintainer / Practitioner subagents + synthesis
- design-brief-template.md — pre-build design doc
- templates/ — SKILL.md and README starters
- scripts/validate-skill.sh — convention checker
Scripts: read scripts/validate-skill.sh before running. Double-check paths and flags every time.
Output: {SKILL_OUTPUT_DIR}/create-a-skill/ — see ../OUTPUT.md
Step 0 — Mode
| Mode | When |
| new | User wants a skill that doesn't exist |
| improve | User names an existing skill in skills/ to extend or fix |
| from conversation | Infer workflow from current chat — confirm before building |
Locate repo root: skills-i-use/ (this repo) or ask user for path.
Step 1 — Intake (hard questions)
Read questions.md. Ask required + hard questions in one message.
Do not proceed until required items are answered. Push back on weak answers — use follow-ups from the red-flag table.
Scan existing skills before intake completes:
ls skills/For each existing skill, read description in frontmatter (first 20 lines of SKILL.md).
Note overlap in the brief.
If user provided verbatim wording for the skill, preserve it exactly in output files.
Step 2 — Build Skill Brief
Compile answers into:
WORKFLOW: [repeatable steps]
FREQUENCY: [how often used]
TRIGGERS: [phrases]
OUTPUTS: [artifacts + skill-outputs path]
STOP_CONDITION: [done when]
MODE: [new | improve]
TARGET: [skill name or existing folder]
OVERLAP: [existing skills]
OUT_OF_SCOPE_V1: [explicit]
TRUST_BOUNDARY: [commit/push/delete/spend — yes/no]
VERBATIM: [any exact copy from user]Step 3 — Reviewer Subagents (parallel)
Read reviewers.md. Run Skeptic, Maintainer, and Practitioner in parallel with the skill brief.
Then run Synthesis (fourth subagent or orchestrator) → DECISION: BUILD NEW | EXTEND | REJECT.
If REJECT — explain why and suggest alternative (script, rule, extend X). Stop.
If EXTEND — read existing skill folder fully before writing. Prefer minimal diff.
Step 4 — Design Brief
Fill design-brief-template.md.
Path: {output_dir}/design-YYYY-MM-DD-[skill-name].md
Present brief to user. Ask: "Proceed to write files?" — stop on no.
Step 5 — Write Skill Files
New skill
Create skills/[skill-name]/ using templates/ as starters. Replace {{placeholders}}.
Typical layout per CONTRIBUTING.md:
skills/[skill-name]/
├── SKILL.md
├── README.md
├── [principles.md | scripts/ | templates/ as needed]Improve existing
- Read entire existing folder
- Apply changes from brief — do not duplicate a new folder unless synthesis says split
- Preserve working triggers unless intentionally breaking (note in brief)
- Extract bloat from SKILL.md into new supporting files if over ~250 lines
Rules while writing:
- Agent-agnostic — subagents/shell, not vendor APIs
- Output →
{SKILL_OUTPUT_DIR}/[skill-name]/ - Safe defaults — no auto-commit/push
- Third-person
descriptionwith what + when
Step 6 — Validate
bash skills/create-a-skill/scripts/validate-skill.sh skills/[skill-name]Fix all FAIL| items. Address WARN| items or document why waived in design brief.
Step 7 — Register in Repo
Update registry files (skip if improving and row already exists):
- README.md — add row to skills table (alphabetical or logical order)
- skills/OUTPUT.md — add subfolder to layout tree if skill produces output
- Cross-skill section in OUTPUT.md — if skill reads/writes another skill's index
Update {output_dir}/index.md:
# Create-a-Skill Designs
| Date | Skill | Decision | Brief |
|------|-------|----------|-------|Step 8 — Ship (user confirms)
- List all created/changed files
- Show validation result
- Suggest
@ship-checkif scripts were added - Ask: "Commit to repo?" — only on explicit yes
git add skills/[skill-name]/ README.md skills/OUTPUT.md
git commit -m "feat: add [skill-name] skill"Ask separately before push. Never push without explicit request.
Edge Cases
- Skill outside skills-i-use — write files where user points; skip Step 7 registry
- Meta: improving create-a-skill itself — same workflow; extra care on regressions
- User says "just write it" — still run validate; abbreviated intake OK if context is rich
- Duplicate skill name — fail validation; pick new name with user
- Subagents unavailable — orchestrator plays all three reviewers sequentially
Invocation Examples
@create-a-skill I want a skill for retrieval benchmarks
improve ship-check — add Python project support
create a skill from this conversation
new skill: legal contract review with persona debate
author a skill, ask me hard questions firstDoes 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.

