Skills library · How-to
A skill file is a reusable ChunkMaps method pattern in Markdown. It tells a human or agent when to use a pattern, what inputs to gather, which steps to follow, what output shape to produce, and which guardrails to enforce. Articles teach; skills are executable playbooks.
Where the library lives
| Location | Role |
|---|---|
skills/ (repo root) | Source of truth - edit here |
TEMP/skills/ | Web preview mirror - copied on every rebuild |
Do not edit TEMP/skills/ expecting changes to stick. Edit repo-root skills/, then rebuild.
Open a skill from the TEMP site
- Open Home or go to Articles & Sitemap.
- Use a Skills library card, or open a mapped article (persuasion / agentic pillars) and click its skill CTA.
- Open the Skills library page (primary index).
- Read the editable how-to source: skills/how-to-use-skill-files.md.
Apply a skill (humans + agents)
- Load the skill - open the matching
.md(or attach it into the agent context). - Confirm fit - check When to use / audience; switch skills if the job does not match.
- Gather inputs - complete the skill's input table before generating answers.
- Follow the procedure - run the numbered ChunkMaps steps in order.
- Produce the output schema - fill the skill template (outline, YAML, tables, CHUNKS_OK, blackboard, etc.).
- Verify guardrails - run failure-mode checks; repair invented evidence, domain drift, or missing citations/IDs.
- Hand off - deliver structured output to the human, pipeline, or next agent.
Load and follow the ChunkMaps skill: <skill-file-name>.md Use only the procedure and schemas in that skill. Gather missing inputs by asking concise questions. Return the skill's output template, then list guardrail checks you performed.
Update or add a skill
- Edit or create
skills/chunkmap-<slug>.md(YAML frontmatter + standard sections). - Update the table in
skills/README.md. - Optionally set
skill: {file, label, blurb}on a TEMP page in_build_temp_site.py. - Rebuild with
python _build_temp_site.py. - Bump the skill
versionwhen behaviour or output schema changes.
Which skill to pick
| Skill | Pick when you need to... |
|---|---|
| Persuasion Flow (raw .md) | Structure human persuasion: Chunk Map, Journey of Persuasion, counterarguments, CHUNKS_OK |
| Agent Blackboard | Run multi-agent swarms with a shared state schema and DAG prompt-trees (anti-drift) |
| Policy to Pipeline | Compile unstructured policy/SOP prose into versioned, citable logic chunks for AI pipelines |