Categories.seoDescriptions.all
Discover the world's most comprehensive directory for AI Agent Skills and community-driven MCP Servers (Model Context Protocol). Seamlessly find and install high-quality developer integrations for your Claude Desktop, Cursor, or custom AI agents. From cloud operations to API integrations, explore verified open-source MCP practices to supercharge your AI workflows.
autonomous-loops is a skill providing patterns and reference implementations for running Claude Code in autonomous loops. It covers architectures ranging from simple sequential pipelines to RFC-driven multi-agent DAG systems for continuous AI agent operation.
Plankton-code-quality is a write-time code quality enforcement system that integrates formatting, linting, and Claude-powered fixes directly into developer workflows via PostToolUse hooks. It automatically processes code modifications and applies corrections through AI-assisted subprocesses.
agent-harness-construction is a developer skill for designing and optimizing AI agent action spaces, tool definitions, and observation formatting. It focuses on core constraints like action space quality and recovery quality to achieve higher completion rates for agents like Claude.
agentic-engineering is an AI agent skill for engineering workflows that uses an eval-first execution loop, decomposes work into agent-sized units, and routes model tiers by task complexity to enforce quality and risk controls.
ai-first-engineering is an operating model for development teams where AI agents generate a significant portion of code implementation. It focuses on process design, review methodologies, and architecture requirements specifically optimized for AI-assisted development workflows with tools like Claude Code.
continuous-agent-loop is a canonical skill for v1.8+ that provides patterns for implementing continuous autonomous agent loops. It features a loop selection flow for different use cases and includes quality gates, evals, and recovery controls for building reliable AI agent systems.
enterprise-agent-ops is a skill for operating long-lived AI agent workloads with enterprise-grade operational controls. It provides capabilities for runtime lifecycle management, observability (logs, metrics, traces), safety controls, and change management beyond single CLI sessions.
nanoclaw-repl is a zero-dependency, session-aware REPL built on Claude for operating and extending the NanoClaw v2 AI agent. It features persistent markdown-backed sessions, dynamic skill loading, session branching, history compaction, and export capabilities to MD, JSON, and TXT formats.
ralphinho-rfc-pipeline is an AI agent skill that implements an RFC-driven multi-agent DAG execution pattern. It breaks down large features into smaller, managed work units, orchestrating them through stages including decomposition, unit assignment, implementation, validation, and integration via a merge queue.
babysit-pr is a CLI-based AI agent that automates the monitoring of a GitHub Pull Request. It runs persistently in your terminal, tracking CI success, review comment status, and merge conflicts, and only exits upon PR merge/close or when user intervention is required.
e2e-testing is a comprehensive testing approach that utilizes Playwright patterns to ensure stable and efficient end-to-end testing for AI agents and applications
setup-dev is a skill for setting up a local development environment for error tracking and performance monitoring using Sentry.