$npx -y skills add Archive228/loopkit --skill templateOne sentence that describes WHEN to invoke this skill. Write it as a trigger phrase, not a summary — include the words a user would say ("fix bug", "add feature", "review diff"). The agent routes on this field.
| 1 | # My New Skill |
| 2 | |
| 3 | <!-- |
| 4 | Anthropic-shaped SKILL.md template. Copy this folder to skills/<your-name>/ |
| 5 | and edit. Keep the body under ~150 lines — long bodies are a context tax on |
| 6 | every turn where this skill is loaded. |
| 7 | |
| 8 | Recommended structure: |
| 9 | 1. One-paragraph statement of the failure mode this skill exists to prevent. |
| 10 | 2. A short checklist or numbered procedure the agent should follow. |
| 11 | 3. Red flags — the shortcuts an agent takes when it wants to skip this. |
| 12 | 4. Evidence — a required before/after pair proving the skill actually |
| 13 | changes behaviour. Ships in the sibling evidence/ folder. |
| 14 | 5. (optional) `references/` sibling files with deeper material the skill |
| 15 | lazy-loads only when needed. |
| 16 | |
| 17 | Do not include: |
| 18 | - Marketing copy. |
| 19 | - Duplicates of other skills (link, do not copy). |
| 20 | - Project-specific rules — those belong in .claude/rules/. |
| 21 | |
| 22 | Frontmatter contract (validated by scripts/validate-skills + CI): |
| 23 | - `name` — kebab-case, unique across the repo, matches the directory name. |
| 24 | - `description` — one non-empty line, phrased as a trigger. |
| 25 | - Anything else is optional and not currently enforced. |
| 26 | --> |
| 27 | |
| 28 | Without this skill, the agent tends to <failure mode>. This skill enforces <the alternative>. |
| 29 | |
| 30 | ## Steps |
| 31 | |
| 32 | 1. <step> |
| 33 | 2. <step> |
| 34 | 3. <step> |
| 35 | |
| 36 | ## Red flags — STOP |
| 37 | |
| 38 | | Thought | Reality | |
| 39 | |---|---| |
| 40 | | "I've done this before, I don't need the checklist" | Skills evolve; re-read the current SKILL.md. | |
| 41 | | "This case is simple" | Simple cases become complex mid-task. | |
| 42 | | "The user didn't ask for this rigor" | The user asked for a working result. Rigor is how you get one. | |
| 43 | |
| 44 | ## Evidence |
| 45 | |
| 46 | Every skill ships with a real before/after pair proving it changes agent |
| 47 | behaviour on a concrete task. Without this, the skill is a claim, not a |
| 48 | tool. |
| 49 | |
| 50 | - `evidence/before.md` — a verbatim snippet of what the agent does on a |
| 51 | representative task WITHOUT this skill loaded. Include the prompt, the |
| 52 | agent's response, and what went wrong (or what got skipped). |
| 53 | - `evidence/after.md` — the same prompt with this skill loaded. Show the |
| 54 | changed behaviour and why it is better. |
| 55 | |
| 56 | If you cannot produce a before/after pair, the skill is not ready to ship. |
| 57 | Land it as a draft in a branch, gather the evidence, then merge. |
| 58 | |
| 59 | ## References |
| 60 | |
| 61 | - `references/<topic>.md` (only if needed; keep the body lightweight) |