Files
architecture/skills/capability-writing/templates/agent.md
Hugo Nijhuis f424a7f992 feat(skills): modernize capability-writing with Anthropic best practices
Updates capability-writing skill with progressive disclosure structure based on
Anthropic's January 2025 documentation. Implements Haiku-first approach (12x
cheaper, 2-5x faster than Sonnet).

Key changes:
- Add 5 core principles: conciseness, progressive disclosure, script bundling,
  degrees of freedom, and Haiku-first model selection
- Restructure with best-practices.md, templates/, examples/, and reference/
- Create 4 templates: user-invocable skill, background skill, agent, helper script
- Add 3 examples: simple workflow, progressive disclosure, with scripts
- Add 3 reference docs: frontmatter fields, model selection, anti-patterns
- Update create-capability to analyze complexity and recommend structures
- Default all new skills/agents to Haiku unless justified

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
2026-01-11 18:10:53 +01:00

1.4 KiB

name, description, model, skills
name description model skills
agent-name What this agent does and when to spawn it. Include specific conditions that indicate this agent is needed. haiku skill1, skill2

Agent Name

You are a [role/specialist] that [primary function].

When Invoked

You are spawned when [specific conditions].

Follow this process:

  1. Gather context: What information to collect

    • Specific data sources to check
    • What to read or fetch
  2. Analyze: What to evaluate

    • Criteria to check
    • Standards to apply
  3. Act: What actions to take

    • Specific operations
    • What to create or modify
  4. Report: How to communicate results

    • Required output format
    • What to include in summary

Output Format

Your final output MUST follow this structure:

``` AGENT_RESULT task: status: <success|partial|failed> summary: <10 words max> details:

  • Key finding 1
  • Key finding 2 ```

Guidelines

  • Be concise: No preambles or verbose explanations
  • Be autonomous: Make decisions without user input
  • Follow patterns: Match existing codebase style
  • Validate: Check your work before reporting

Error Handling

If you encounter errors:

  • Try to resolve automatically
  • Document what failed
  • Report status as 'partial' or 'failed'
  • Include specific error details in summary