crystools-skills
By crystian
Create structured SKILL.md templates for new Claude Code AI agent skills with triggers, steps, and guardrails, then analyze executions via conversation friction, file diffs, feedback, and diagnostics to generate concrete improvements for better efficiency.
npx claudepluginhub crystian/skillsComponent Overview
Component Details
Skills (2)
Third-person description of what this skill does, when it should activate, and how to invoke it. Include concrete trigger phrases the agent can match against. Max 1024 characters. Example: "Reviews SQL migrations for destructive operations and suggests safer alternatives. Activates when the user asks to review a migration, or runs `/crystools-skills:your-skill-name`."
Optimizer that refines and professionalizes AI agent skills through real usage — saves tokens, eliminates redundancy, and tightens instructions so skills cost less to run. Learns from mistakes, reviews quality, and improves over time. Observes skill execution in the current conversation, analyzes up to four sources (conversation friction, file diffs, user feedback, static diagnostic) plus accumulated lessons, and proposes concrete improvements to the target skill's SKILL.md. Works with Claude Code and compatible SKILL.md-based agent frameworks. Use after executing any skill: `/skill-optimizer [name]` or `/skill-optimizer` to auto-detect. `--review` processes accumulated lessons.
README
crystools-skills
Open collection of AI agent skills — reusable, framework-agnostic SKILL.md packages for Claude Code and compatible agents.
Table of Contents
What's a skill?
A skill is a self-contained directory with a SKILL.md file. It packages "how to do X" in a way any SKILL.md-aware agent can execute, with no shared runtime and no global state. Copy it, fork it, move it between projects — it just works.
Skills follow a three-layer progressive disclosure model: lightweight frontmatter (for activation), a bounded body (for execution), and optional references (loaded on demand). See HUMAN.md for the design philosophy.
Available skills
| Skill | Description | Docs |
|---|---|---|
| skill-optimizer | Kaizen for AI agent skills — observes execution, diagnoses friction, proposes concrete SKILL.md improvements | README |
| skill-map | Map, inspect and manage collections of interconnected Markdown skills, agents and docs — graph explorer + CLI + optional LLM layer. Lives in its own repo. | README |
More coming. Want to add one? Jump to Contributing.
Install
# All skills via Claude Code plugin
claude /plugin install crystian/skills
# All skills via npx
npx skills add crystian/skills
# A single skill
npx skills add crystian/skills --skill skill-optimizer
Once installed, invoke with the plugin namespace: /crystools-skills:<skill-name>.
Update
npx skills check # check for updates
npx skills update # update all
npx skills add crystian/skills # reinstall latest
Contributing
Contributions are welcome — new skills, improvements, bug fixes, docs. Start here:
- Read CONTRIBUTING.md.
- For a new skill, open a New skill proposal first to align on scope.
- Copy
skills/_template/as your starting point. - Open a PR; CI validates SKILL.md structure automatically.
Quick start for a new skill:
cp -r skills/_template skills/my-new-skill
# edit skills/my-new-skill/SKILL.md and README.md
Project layout
.claude-plugin/
plugin.json # plugin manifest
.github/
ISSUE_TEMPLATE/ # issue forms
workflows/ # CI (validate + auto-bump version)
pull_request_template.md
scripts/
validate-skills.py # SKILL.md linter (used by CI and locally)
skills/
_template/ # starting point for new skills
skill-optimizer/
SKILL.md # skill definition (frontmatter + body)
README.md # user-facing docs
references/ # optional, loaded on demand
CONTRIBUTING.md # how to contribute
CODE_OF_CONDUCT.md # community standards
SECURITY.md # how to report vulnerabilities
CHANGELOG.md # release notes
HUMAN.md # design notes for humans (not for agents)
CLAUDE.md # instructions for Claude / agents
Documents
- CONTRIBUTING.md — how to contribute, skill standard, PR workflow.
- CODE_OF_CONDUCT.md — community standards (Contributor Covenant v2.1).
- SECURITY.md — vulnerability reporting policy.
- CHANGELOG.md — release history.
- HUMAN.md — design principles and project vision (for humans).
- LICENSE — MIT.
Related projects
- skill-map — companion tool that maps, inspects and manages collections of Markdown-based skills, agents, commands and docs. Graph explorer + CLI + optional LLM layer. Works on any ecosystem of interconnected MDs (not limited to this repo).
License
MIT. See LICENSE.
Similar Plugins
skill-forge
Ultimate Claude Code skill creator. Design, scaffold, build, review, evolve, and publish production-grade AI agent skills following the Agent Skills open standard and 3-layer architecture.
Help us improve
Share bugs, ideas, or general feedback.