description-craft
Skill1 starsChanged 7 months agoMITView on GitHub
Diagnose and rewrite a SKILL.md frontmatter description so Claude reliably triggers the skill at the right moment. Use when a skill fails to trigger, triggers on the wrong tasks, or when writing a new skill and asking "is this description good". Returns a diagnosis, a rewritten description, and why the rewrite fires better.
In this repository11 other files it gives its agents.
- assumption-surfaceSkill
- goal-criteriaSkill
- karpathy-codingSkill
- lean-responseSkill
- simplicity-auditSkill
- skill-compressSkill
- skill-graphSkill
- surgical-diffSkill
- token-auditSkill
- token-budget-checkSkill
- vixero-councilSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 65
Never write descriptions longer than 300 chars.
- Distinctive: line 66
Never start with "This skill" — waste.
- Distinctive: line 67
Never use "helps with" — vague trigger, weak fire.
- Distinctive: line 68
If the skill does more than one thing, recommend splitting it.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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
description-craft
Fix the load-gate. The description is the only part of a skill Claude sees every turn — bad description means the skill never fires.
Trigger on
- "my skill doesn't trigger"
- "is this description good"
- "help me write the description / frontmatter"
- "write a SKILL.md description for X"
What a description must contain
- What it does. Verb + object. "Rewrite verbose skills into lean form."
- When to use it. Explicit trigger signals. "Use when the user asks to compress, shrink, tighten..."
- Trigger keywords. Synonyms the user might actually type — not just the formal name. A skill called
threat-modelshould mention STRIDE, PASTA, threat modeling, risk assessment. - Output shape. One phrase on what the user gets back. "Returns a ranked list of threats with mitigations."
Diagnosis checklist
For the given description, mark each:
| check | pass / fail |
| length 100–300 chars | |
| starts with a verb | |
| contains "Use when" or equivalent | |
| lists 3+ synonym triggers | |
| names the output / return value | |
| no meta-fluff ("This skill will...") | |
| no hedging ("might help with...") |
Three or more fails → rewrite from scratch. One or two fails → patch.
Rewrite pattern
{verb} {object}. Use when the user asks to {synonym1}, {synonym2}, {synonym3}, or {synonym4}. Returns {output-shape}.Output
diagnosis:
length: {n} chars ({pass/fail})
starts with verb: {pass/fail}
has "Use when": {pass/fail}
synonym count: {n} ({pass/fail})
names output: {pass/fail}
fluff: {pass/fail}
rewritten description:
{new description}
why it fires better:
- {reason 1}
- {reason 2}Hard rules
- Never write descriptions longer than 300 chars. Anthropic docs recommend specific and short.
- Never start with "This skill" — waste.
- Never use "helps with" — vague trigger, weak fire.
- If the skill does more than one thing, recommend splitting it. One skill, one job.
- The same skill name must appear nowhere in the description body — that is the
namefield's job.
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.

