M
MCP Directory CURATED DIRECTORY
911 MCP servers indexed
中文
← Back to directory
COMPARE UP TO 4 SERVERS

Compare MCP servers

Compare scores, permissions, risks, and fit in one decision-focused table.

DimensionVolcano Agent SDKBuild AI agents that combine LLM reasoning with MCP tool calls in a few lines of TypeScriptMartinLoop MCP ServerAn independent control layer that gives AI coding agents budgets, brakes, and receipts
FMRS38 / 100 · D79 / 100 · B
Reliability4 / 2012 / 20
Security and permissions8 / 2018 / 20
Maintenance11 / 2016 / 20
Documentation9 / 2018 / 20
Setup experience6 / 2015 / 20
Best for
  • Backend/platform engineers already using TypeScript who need to orchestrate LLM calls and multiple MCP servers within their own application
  • Teams building multi-agent delegation workflows (research, writing, task handling, etc.) with autonomous coordination
  • Teams that need to mix several LLM providers in one workflow with built-in retries, streaming, and tracing
  • Engineering teams scaling AI coding agents while controlling spend
  • Platform teams that need auditable, rollback-aware agent execution records
  • Organizations using multiple agents (Claude, Codex, Gemini) under one governed flow
Not for
  • Users who want an out-of-the-box MCP server that exposes specific business tools (this project provides no tools itself — it requires connecting to external MCP servers)
  • Developers working in Python or other non-JavaScript/TypeScript stacks
  • Non-technical users hoping to add it via a Claude Desktop-style configuration file
  • Developers who just want a simple code-completion helper without run governance
  • Teams unwilling to accept local JSONL run records with HMAC signing
  • Environments without a supported adapter (Claude/Codex/Gemini CLIs, direct provider, or verifier-only)
Required permissions
  • Requires developer-supplied API keys for the LLM provider(s) in use (e.g., OPENAI_API_KEY)
  • Requires network access to connect to whatever MCP server endpoints the developer deploys (e.g., http://localhost:8001/mcp)
  • Because it supports automatic tool selection, it implicitly gains whatever permissions the connected MCP servers grant to their tools
  • Read/write access to the configured repo root (writable scope restrictable via --allow-path/--deny-path)
  • Ability to invoke the chosen agent CLI (Claude/Codex/Gemini, etc.) and its underlying model account
  • Local disk writes for run records, receipts, and share outputs (share/run-receipt., etc.)
  • Execution of user-configured verifier commands (e.g. npm test)
Risks and side effects
  • Automatic tool selection means the LLM decides which tools to invoke based on the prompt; if connected to MCP servers with real side effects (creating tasks, sending requests), there is a risk of unintended actions from prompt injection or model misjudgment
  • In multi-agent delegation mode, the coordinator autonomously decides when the workflow is done, which could lead to unexpected repeated calls or resource use if not designed carefully
  • The README does not describe specific authentication, sandboxing, or permission-isolation mechanisms — the security boundary of connected MCP servers is the developer's responsibility
  • Verifier commands are executed; poorly chosen verifier commands carry security risk despite preflight policy checks
  • Cost and token figures may be estimates (labeled with provenance) and should not be treated as settled accounting
  • Receipt integrity must be verified before use as trustworthy evidence; missing rollback evidence is flagged as EVIDENCE_BOUNDARY
  • Policy defaults come from martin.config.yaml and can be overridden by CLI flags; misconfiguration can loosen safety boundaries
Supported clientsClaude Code, Codex, Gemini CLI, Cursor, VS Code, GitHub Copilot, Continue
Tools00