Menu
Use when creating, updating, or improving agent skills.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
| name | writing-skills-excellence |
| description | Use when creating, updating, or improving agent skills. |
| category | meta |
| risk | unknown |
| source | community |
| date_added | 2026-02-27 |
Dispatcher for skill creation excellence. Use the decision tree below to find the right template and standards.
Create a NEW skill:
Improve an EXISTING skill:
Verify Compliance:
| Component | Purpose |
|---|---|
| CSO | "SEO for LLMs". How to write descriptions that trigger. |
| Standards | File naming, YAML frontmatter, directory structure. |
| Anti-Rationalization | How to write rules that agents won't ignore. |
| Testing | How to ensure your skill actually works. |
references/templates/---
name: my-technique
description: Use when [specific symptom occurs].
metadata:
category: technique
triggers: error-text, symptom, tool-name
---
# My Technique
## When to Use
- [Symptom A]
- [Error message]
| Mistake | Fix |
|---|---|
| Description summarizes workflow | Use "Use when..." triggers only |
| Vague description | Put "Use when..." trigger conditions in description; Claude Code uses this for discovery |
| Generic name ("helper") | Use gerund (creating-skills) |
| Long monolithic SKILL.md | Split into references/ |
See gotchas.md for more.
Before deploying any skill:
name field matches directory name exactlySKILL.md filename is ALL CAPSdescription for discovery, not metadata.triggers)references/ for more)@ force-loading in cross-references/write-skill command for guided skill creationCreate a Tier 1 skill (user-global):
mkdir -p ~/.claude/skills/my-technique
touch ~/.claude/skills/my-technique/SKILL.md
Create a Tier 1 skill (project-local):
mkdir -p .claude/skills/my-technique
touch .claude/skills/my-technique/SKILL.md
Create a Tier 2 skill:
mkdir -p .claude/skills/my-skill/references/core
touch .claude/skills/my-skill/{SKILL.md,gotchas.md}
touch .claude/skills/my-skill/references/core/README.md
Apply Donella Meadows' systems thinking framework to map, diagnose, and redesign any system — organizational, technical, ecological, or policy. Covers stocks/flows, feedback loops, system archetypes, leverage points, and concrete intervention recommendations.
Keep the main conversation clean by forking implementation work to sub-agents. Use when tasks would otherwise flood context with file reads, shell output, build logs, or multi-step execution details.
Run, validate, and manage PAS (Pascal's Discrete Attractor) DOT-based AI pipelines. Use when launching a pipeline end-to-end, resuming interrupted runs, capping spend, or generating pipelines from spec/PRD files.
Manage the Reck software factory — register repos, run AI tasks in containers, schedule background pipelines, and monitor results. Use when provisioning a new task, checking task status, or setting up recurring automation.
Generate acceptance criteria for a user story or feature. Produces testable, specific criteria in Gherkin (Given/When/Then), checklist, or rules-based format with a completeness check. Use when writing or reviewing AC for any story, ticket, or feature spec.
Use when implementing security-critical code (auth, crypto, secrets, TLS, CORS), financial arithmetic, protocol parsers, deserialization of untrusted data, or any code the user wants to be "production-ready," "correct," or "verified." Requires citing authoritative sources — RFCs, NIST, OWASP, official library docs — before writing, not after.