MCP servers that give coding agents real developer tooling: linters, package registries, test runners, language docs and local build commands. 13,958 servers in this category.
A Python runtime for tool-calling agents that supports multiple LLM providers and provides streaming, session management, and JIT tool loading.
Provides guardrails and project maps for AI coding agents to prevent context drift, token waste, and destructive refactoring.
A lightweight Python framework for building Model Context Protocol servers with tools, resources, and prompt templates for Claude.
Performs static analysis and drift detection on MCP servers to classify tool blast radius and detect prompt injection vulnerabilities.
Provides an authorization boundary and security regression testing layer for AI agents and MCP tools to prevent unauthorized tool execution.
An agent-native control plane for managing post-training workflows including dataset validation, supervised fine-tuning, RLHF, and model evaluation.
Runs a comprehensive trust and safety audit on MCP servers using static analysis, fuzz testing, and output fidelity checks.
Provides a provenance and audit layer for AI agents to record runs, track tool calls, and validate comparison results on research infrastructure.
A pytest plugin that asserts MCP server configuration profiles expose only the intended tools and no unauthorized capabilities.
Acts as a security gateway for MCP servers to block SSRF, secret leaks, prompt injection, and cross-tool exfiltration.
A fast Python REST framework with a Rust core that exposes routes to AI agents via the Model Context Protocol.
Transforms research papers into software requirement specifications, development checklists, and functional code using pluggable AI providers.
Tests MCP servers using automated experiments in Docker sandboxes to record and analyze JSON-RPC wire events.
Provides a control plane for MCP tools featuring authorization, chat-context isolation, audit logging, and managed execution environments.
Analyzes MCP tool surfaces to assess token costs, capability risks, authentication posture, and configuration hygiene.
Enables AI agents to analyze, edit, and automate Win32 PE resource manipulation through a secure, high-performance workbench and MCP server.
Provides a self-hosted remote workspace for AI agents featuring sandboxed shell execution, file management, and memory persistence.
Materializes downstream MCP tool results into an in-memory DuckDB database so agents can perform exact SQL aggregation over tool outputs.
Manage MCP server configurations and tool permissions for Claude Code and Codex via a local CLI.
Provides a synthetic multi-server demo to demonstrate enterprise AI patterns, tool failures, and guardrail implementation using airline booking data.