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.

DimensionAsk Ollama MCP ServerBridge Claude with local Ollama models for private AI-to-AI collaboration — no API keys, fully local.MartinLoop MCP ServerAn independent control layer that gives AI coding agents budgets, brakes, and receipts
FMRS54 / 100 · D79 / 100 · B
Reliability8 / 2012 / 20
Security and permissions11 / 2018 / 20
Maintenance12 / 2016 / 20
Documentation12 / 2018 / 20
Setup experience11 / 2015 / 20
Best for
  • Teams or individuals with strict data-privacy requirements whose code cannot leave the machine
  • Developers already running Ollama locally who want to reuse that compute
  • Claude Code, Cursor, or Codex CLI users who want a second opinion from a different model
  • Users who want to avoid per-token API costs
  • 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
  • Environments without Ollama installed or running
  • Workflows needing large cloud-model capability or very large-context reads
  • Users wanting multi-provider parallel fan-out (multi-llm) or cross-provider orchestration — use the unified @ask-llm/mcp instead
  • Users seeking official vendor support or an official relationship
  • 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
  • Local process communication with the MCP client over stdio
  • Access to the local Ollama HTTP service (default http://localhost:11434, configured via OLLAMA_HOST)
  • Reads prompt content passed by the MCP client (the code or text sent to local Ollama)
  • 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
  • Prompt content is sent to the local Ollama service; if OLLAMA_HOST is misconfigured to a remote host, data could leave the machine
  • Local models are generally weaker than frontier cloud models, so reviews may be incomplete or produce false positives
  • If the requested model has not been pulled locally, the call fails with an ollama pull error
  • Third-party, unofficial project — not endorsed by Anthropic, Google, OpenAI, or xAI
  • Long local inference may exceed the default 5-minute timeout (GMCPT_TIMEOUT_MS) and be interrupted
  • 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 Desktop, Claude Code, Cursor, CodexClaude Code, Codex, Gemini CLI, Cursor, VS Code, GitHub Copilot, Continue
Tools30