Agent Skills

Skills for your AI agents

Free, production-tested skills in SKILL.md format. Drop one into Claude Code (or any agent that reads markdown instructions) and it gains a repeatable, expert workflow — distilled from real work shipping AI products.

Plain markdown — no dependenciesOne-command installFree to use and adapt

AI Engineering

LLM Cost Audit

llm-cost-audit.md

Find where your AI spend is leaking — caching wins, model-tier mismatches, and token waste.

LLMCost OptimizationCachingObservability

AI Engineering

Agentic Error Recovery

agentic-error-recovery.md

Harden AI agents for production — idempotency, loop guards, fallbacks, and real observability.

AI AgentsReliabilityIdempotencyProduction

AI Product

AI Feature Rollout

ai-feature-rollout.md

Ship AI features behind flags, guardrails, and kill switches — not big-bang releases.

Feature FlagsA/B TestingRolloutAI Product

E-commerce

Shopify App Review

shopify-app-review.md

Catch App Store rejections before Shopify does — OAuth, webhooks, billing, and listing audit.

ShopifyComplianceApp StoreCode Review

Backend

Prisma + Postgres Performance

prisma-postgres-performance.md

Find the queries that will hurt at 10× traffic — N+1s, missing indexes, and over-fetching.

PrismaPostgreSQLPerformanceDatabase

How agent skills work

Pick a skill

Each skill is a single SKILL.md file — an expert workflow written as instructions an AI agent can follow reliably.

Install it

Copy it into your agent's skills directory (for Claude Code: ~/.claude/skills/) or paste it straight into a conversation.

Invoke it

Ask for the task the skill covers — the agent applies the workflow, checklists, and output format automatically.