# mgechev/skills-best-practices

Write professional-grade skills for agents, validate them using LLMs, and maintain a lean context window.

Repository: https://github.com/mgechev/skills-best-practices
Canonical: https://ross.abutalabs.com/products/skills-best-practices
Language: Python
License Family: other
Last push: 2026-07-27T16:27:51+00:00

## Health v2 (maintenance only)
Score: 57/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 94, release rhythm 35, longevity 13
- inputs: {"age_days": 190, "days_push": 37, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2237, forks 166 (observed 2026-08-28T04:06:29.387937+00:00)

## What it is
A guide and reference repository of best practices for writing professional-grade agent skills (SKILL.md files) for LLM agents like Claude. It covers skill directory structure, frontmatter optimization for discoverability, and progressive disclosure to keep the agent's context window lean.

## Use cases
- how to write agent skills for claude
- best practices for SKILL.md files
- optimize skill descriptions for agent triggering
- keep agent context window lean with progressive disclosure
- structure a skill with scripts references and assets
- validate agent skills with LLMs
- naming conventions for skill frontmatter

## When to choose
- you are authoring or maintaining skills for Claude or similar LLM agents
- you want a concise, opinionated checklist for skill structure and discoverability
- your skills load too much context and you need progressive-disclosure patterns

## When to avoid
- you need comprehensive official documentation rather than a concentrated guide
- you are building agent frameworks or runtimes rather than writing skills
- you need a tool that grades skills - use the companion skillgrade repo instead

## Facets
- artifact type: learning-resource
- maturity: active
- function: documentation, prompt-engineering, agent-framework, developer-tools
- domain: artificial-intelligence, developer-tools, tutorials, documentation
- platform: python, cli, cross-platform
- tags: agent-skills, claude-skills, best-practices, context-window, llm-validation, skill-authoring, ai-agents

## Member repositories
- mgechev/skills-best-practices (main) score 57

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:29.387937+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T02:44:15.919955+00:00, confidence not recorded.
  - readme: https://github.com/mgechev/skills-best-practices (fetched 2026-08-28T04:06:29.387937+00:00, sha 1ba82c4fbb7e)
- Data as of 2026-08-30T08:39:29.467469+00:00.