This repository is English-only and Markdown-only.
| Type | Description |
|---|---|
| New prompts | Agent, project-type, or workflow prompts |
| Improvements | Better structure, clarity, outcome quality |
| Examples | Real-world APEI walkthroughs |
| Maintenance | Link fixes, catalog cleanup, archive actions |
Every prompt must contain:
## Role## Protocol / Core Loop## Phases## Remember(final section)
Every recommendation should end with a concrete:
- decision, or
- tool, or
- validation step.
If a prompt is low-value or overlapping:
- Classify as keep / merge / archive
- Move to
prompts/english/agents/archive/ - Update active indexes and README
- Add rationale to archive index
- Markdown renders correctly
- No spelling or grammar errors
- Internal relative links resolve
- Hypothetical example paths are plain code literals (not Markdown links)
- Catalog/index entries are updated
-
llms.txtis updated when primary navigation or core prompts change - No vague advice language
Optional local checks:
grep -r '\[.*\](.*\.md)' prompts/ | head
npx markdownlint-cli2 '**/*.md'feat: add new prompt
fix: correct catalog link
docs: improve usage guidance
update: archive overlapping prompt