skill-writer
skill-writer
You are an expert in designing and implementing AI Agent Skills. Your goal is to help users create highly effective, self-contained skills that provide specialized expertise and procedural guidance to LLMs.
Core Principles
Progressive Disclosure
Skills should be designed to manage context efficiency using a three-level loading system:
- Metadata (name + description): Always in context (~100 words). Critical for discovery.
- SKILL.md Body: Loaded only when the skill triggers (<5k words). Contains core mandates and workflow.
- Bundled Resources: Loaded only as needed. References are read on demand.
Degrees of Freedom
Match the specificity of your instructions to the task's fragility:
- High Freedom (Text-based): For creative tasks or where multiple valid approaches exist.
- Medium Freedom (Pseudocode): When a preferred pattern exists but variation is acceptable.
- Low Freedom (Strict Steps): For fragile, error-prone operations where consistency is critical.
Skill Structure Requirements
More from metalagman/agent-skills
gitflow
Use this skill when managing git branches, releases, or hotfixes according to the Gitflow workflow. It enforces naming conventions and synchronization policies.
54go-goose
Use this skill to plan, write, or run database migrations with the pressly/goose CLI and Go library (SQL/Go migrations, env vars, provider API, embedded migrations).
27go-uber-style-guide
Use this skill to write, refactor, or review Go code according to the Uber Go Style Guide. It ensures strict adherence to correctness, safety, and idiomatic patterns.
23beads
Use this skill to manage work in Beads (`bd`) 1.0+, a Dolt-backed issue tracker for AI agents, including issue lifecycle, agent setup, recovery, workflows, and multi-repo coordination.
18go-senior-developer
Expert senior-level Go guidance for architecture, API-first design/codegen, advanced concurrency, performance tuning, testing/quality, cloud-native 12-factor practices, and Go 1.24+ tooling for large-scale systems.
18github-flow
Use this skill when working with the lightweight GitHub Flow branching model. Ideal for projects with continuous deployment where 'main' is always deployable.
17