| name | author-skill |
| description | Use `author-skill` to get Skill File Authoring when you need to review/write skill files for agents. |
Skill File Authoring
Use this Skill Template to review/write skill files:
Skill Template
---
name: [NAME]
description: Use `[NAME]` to get [TOPIC] if [CONDITION].
---
# [TOPIC]
[TRIGGERS]
---
[CONTENT]
---
[RULES]
Write Skill Template as follow:
- Replace [NAME] with short skill name as follows:
- 4 words max
- Prefix with
author- for skill related to human readable content, articles
- Prefix with
code- for skill related to project source code, configurations, scripts
- Prefix with
design- for skill related to project designs (architecture)
- Replace [TOPIC] with short topic in Caveman English as follows:
- Unique header summarizing topic skill addresses - What knowledge and/or ability agent should expect (e.g. "Deploying Spring Boot Applications")
- Must be < 7 words
- Use exact phrase and case consistently
- Replace [TRIGGERS] with description of when topic is relevant (< 20 words) like "Understand how to build and package Spring Boot Application for production deployments, ..."
- Replace [CONTENT] with translated content from user request into Caveman English:
- Preferably format as structured sequence lists or bullet points
- May include formatted examples when applicable
- Prefer linking to templates, scripts, references, external sources instead of embedding verbose content to keep SKILL.md lean
- Replace [RULES] with important rules regarding agents behavior to apply skill:
- Optional: Only include if behavioral changes are required
- Must be last section SKILL.md
- Only include very critical instructions or limitations that always apply
- When rules contradict: Add conditions when each contradicting rule applies
- Name [TRIGGERS], [CONTENT], [RULES] sections with H2 headers (max 4 words each)
Rules
- ALWAYS write skill files in
.agents/skills/{name}/SKILL.md
- Place templates, scripts, references in subdirectories from skill directory
.agents/skills/{name}/ and link to it from SKILL.md, for example: templates/email_template.html, scripts/process_data.py
- Keep templates, scripts, reference files also lean (only keep important info and valuable example snippets).
- No repetitions
- Only use emojis to highlight important aspects to LLM, like attention, warning, checklists, correct vs wrong
- ALWAYS keep md < 400 lines by:
- Cleaning up redundant content
- Removing excessive examples
- Summarize trivial info in [CONTENT] section
- Only if skill's [CONTENT] section is > 200 lines:
- Divide content into smaller < 200 line sections
- Divided skills user content should not overlap
- Move subdivided content to reference files (located in
.agents/skills/{name}/) and link to it
- As last resort: Reducing/grouping obvious instructions agent can derive by itself
Use this same skill text as an example on how to format other skills.