I built the article-writing skill to turn CLI conversations into publishable markdown articles.
V1 was mostly Codex-oriented, then I iterated: more generic extraction, cleaner deliverables, and 4 tones/voices.
Here is a reproducible method to create a custom skill.
Preconditions#
Having a CLI tool like claude code, copilot, codex or any other with a SKILL.md file to host your custom skill.
SKILLS/<skill-name>/SKILL.mdExample:
SKILLS/article-writing/SKILL.mdHow-to#
1) Define the scope#
Define the skill behavior in one complete sentence:
- what it takes as input,
- what it does,
- what it produces.
Example:
- Input: a raw CLI conversation.
- Action: extract useful messages, then structure them.
- Output: a publishable markdown article.
2) Write minimal frontmatter#
---
name: article-writing
description: Converts an agent conversation into a publishable markdown article.
---3) Build an executable workflow#
## Workflow
1. Identify input
2. Extract useful data
3. Validate intermediate result
4. Ask required parameters
5. Generate final deliverable4) Iterate with small changes#
Simple loop:
- Change one point
- Test on a real case
- Check for regressions
Real example:
- V1: initial flow
- V2: more generic extraction
- V3: deliverables + tones/voices
5) Invoke the skill#
You have two useful modes.
Explicit invocation (recommended at first):
@SKILLS/article-writing/SKILL.mdHere you force usage of that specific skill.
Intent-based invocation (once the skill is well described):
Write an article from our previous conversation.In this case, the agent selects the skill if the SKILL.md description and triggers match.
Practical check: if the response does not follow the expected workflow, invoke the skill explicitly again via @SKILLS/.../SKILL.md.
Resources#
Skill article-writing available here: SKILL.md