Custom AI agent skills for real engineering work — not vibe coding.
Works with any agent that reads markdown: Crush, Claude Code, Cursor, Windsurf, GitHub Copilot, and others.
git clone https://github.com/ericdahl/crush-skills ~/.crush-skills
cd ~/.crush-skills
./scripts/link-skills.shThe script asks which editors to install into, then symlinks each skill into the right directory:
| Editor | Skills directory |
|---|---|
| Crush | ~/.config/crush/skills/ |
| Claude Code CLI | ~/.claude/skills/ |
| Cursor | ~/.cursor/rules/ |
| Windsurf | ~/.codeium/windsurf/memories/ |
| GitHub Copilot Chat | ~/.github/copilot/skills/ |
Or pass editors directly:
./scripts/link-skills.sh crush claudeEach skill is a SKILL.md file. When you invoke the skill name (e.g. /tdd, /diagnose) in your agent, it reads the file and follows the instructions inside. Skills compose — many reference sibling skills.
Invoking a skill:
- Crush / Claude Code:
/skill-name - Cursor / Windsurf:
@skill-nameor reference by name in chat - Any agent: paste or attach the
SKILL.mdcontent, or instruct the agent to read it
Skills created in-house.
| Skill | Description |
|---|---|
crush-code |
Autonomous issue-driven dev loop (TDD, auto-merge PRs, PR health) |
github-triage |
Label-based GitHub issue triage state machine |
| Skill | Description |
|---|---|
find-skills |
Discover and install new agent skills |
| Skill | Description |
|---|---|
copywriting |
Marketing copy for landing pages, features, pricing |
frontend-design |
Production-grade UI with bold aesthetic direction |
landing-page-copywriter |
High-converting copy (PAS, AIDA, StoryBrand) |
web-design-guidelines |
UI/UX + accessibility audit |
| Skill | Description |
|---|---|
coolify-manager |
Manage and troubleshoot Coolify deployments |
daily-devlog |
Gather today's commits/PRs and write a devlog entry |
Skills adopted from other open-source skill repos. Credit to the original authors.
From mattpocock/skills
| Skill | Description |
|---|---|
diagnose |
Disciplined debug loop: reproduce → minimize → fix → regression test |
grill-with-docs |
Stress-test a plan against CONTEXT.md and ADRs |
improve-codebase-architecture |
Find refactoring opportunities informed by domain language |
setup-matt-pocock-skills |
Bootstrap agent skill config (issue tracker, labels, domain docs) |
tdd |
Red-green-refactor TDD loop |
to-issues |
Break a plan/PRD into independently-grabbable GitHub issues |
to-prd |
Turn conversation context into a published PRD |
triage |
Role-driven issue triage state machine |
zoom-out |
Higher-level perspective on unfamiliar code |
caveman |
Ultra-compressed communication mode (~75% fewer tokens) |
grill-me |
Relentless questioning to stress-test a plan |
handoff |
Compact conversation into a handoff doc for the next agent |
prototype |
Throwaway prototype to flesh out a design |
write-a-skill |
Create new agent skills with proper structure |
| Skill | Description |
|---|---|
vibe-check |
Production resilience audit for AI-generated code |
vibe-explain |
Cognitive debt map — surfaces code you don't fully understand |
vibe-guard |
Full safety check: resilience + security + comprehension |
vibe-secure |
Security audit for AI-generated code |