Agent Skills by thomas
Showing 32 vetted skills indexed across 1 GitHub repositories.
project-guidelines-example
Defines a project-specific skill template with YAML frontmatter and structured sections.
ui-ux-pro-max
Search UI/UX style guides with BM25 to surface design patterns and guidelines.
python-testing
Guide Python testing with pytest-driven TDD, fixtures, mocks, and coverage.
docker-patterns
Automate Docker and Docker Compose best-practice patterns for local development and multi-service workflows.
iterative-retrieval
Refine file selection across four phases for subagent context.
postgres-patterns
Apply PostgreSQL indexing, schema, RLS, and upsert patterns for performance and security.
coding-standards
Defines coding standards for TypeScript, JavaScript, React, and Node.js projects.
eval-harness
Automate formal evaluation workflows for Claude Code sessions with pass@k metrics.
cost-aware-llm-pipeline
Route LLM tasks to cheaper models and track costs with immutable records.
prompt-optimizer
Analyze draft prompts and output ready-to-paste optimized prompts.
claude-api
Automates Claude API usage and integration patterns in Python and TypeScript for building Claude-powered applications.
agent-harness-construction
Design agent action spaces and tool definitions with deterministic outputs.
deep-research
Compiles cited reports from multi-source web research using firecrawl and exa MCP tools.
backend-patterns
Design scalable Node.js backend architectures for REST and GraphQL APIs.
api-design
Standardize REST API design including naming, pagination, filtering, error responses, versioning, and rate limiting.
security-scan
Scan Claude Code configuration files for security vulnerabilities using AgentShield.
deployment-patterns
Creates deployment strategies for Dockerized CI/CD pipelines with health checks and rollback plans.
e2e-testing
Automate Playwright end-to-end testing workflows with POM utilities and configuration guidance.
agentic-engineering
Coordinate AI-driven engineering tasks with eval-first validation and tiered model routing.
brainstorming
Organize raw ideas into actionable design concepts with structured inquiry.
content-hash-cache-pattern
Cache file processing results using SHA-256 content hashes as keys.
continuous-learning-v2
Convert Claude Code sessions into atomic instincts with confidence scoring.
verification-loop
Automate end-to-end verification of Claude Code sessions across build, type, lint, test, security, and diff review phases.
database-migrations
Plan reversible database migrations across PostgreSQL, MySQL, and ORMs.