MCP servers that ship as code — a repository or a package you install and run locally (usually over stdio). They have no hosted endpoint, so we cannot probe or grade them: instead each one shows its repository, install command, license, stars and when it was last updated.
Collected from GitHub repositories and npm, PyPI, Cargo, Go, Maven, NuGet and RubyGems packages tagged as MCP servers, the official MCP Registry, Docker's MCP catalog, ModelScope, Baidu MCP World, Cline and other public directories; one entry per repository. A server that also offers a hosted endpoint is listed under Remote instead. These entries are not counted in our reachability, verification or readiness statistics.
-
SuperBryn MCP server — run the voice-agent eval loop and read your production call logs from Claude Code, Claude Desktop, or Codex.
npm MIT -
Your phone, for your AI agents: SMS, OTP codes, contacts and call log for Claude, Hermes, Codex and any MCP client. E2E encrypted, self-hosted relay.
MIT -
Logs, traces and LLM calls from every run, live in VS Code: terminals, debug sessions, Docker and OpenTelemetry, next to production in Datadog, CloudWatch, Loki, Elasticsearch, Cloud Logging and SQL databases. Shared with Copilot, Claude Code or Codex over MCP.
MIT ★ 2 -
Open source phone MCP server for Claude and ChatGPT: let an AI agent place calls and send SMS through Twilio with plan then confirm codes, dry run by default, number allowlists, calling-hour guards and an audit log. For developers and agencies building AI agents that use the phone. Python, self-hosted, MIT.
MIT ★ 2 -
HISTOR — public transparency log of MCP tool definitions: reads what remote MCP endpoints in the official registry advertise (never calls a tool), signs it as MTL/1 labels, appends them to an RFC 9162 Merkle log, and answers /check: is what a client received what was observed? Never calls a server safe.
pypi MIT -
A permission layer in front of MCP tools, and a log that can prove why a call was allowed. Naive: 7 calls execute with no grant, 16 decisions cannot be reconstructed. Guarded: 0 and 0. Offline, zero deps, MIT.
MIT -
aliyun-sls-mcp-servergithub.com/LilMosey/aliyun-sls-mcp-serverupdated this year
阿里云日志MCP,让你的AI Agent帮你分析线上报错!支持Agent侧并发调用工具!Alibaba Cloud Log MCP: Let your AI Agent analyze production errors for you! Supports concurrent tool calls on the agent side.
★ 3 -
mcp-ops-orchestratorgithub.com/akhileshthokala/mcp-ops-orchestratorupdated this year
MCP ops orchestrator with tool-calling, ticket creation, email drafting, structured logs, and human approval gates.
-
Holds risky AI agent tool calls for a person, ties each approval to one exact request, and logs every decision. 210 checks, synthetic data.
MIT -
AI-agent tool-call policy gateway with least privilege, parameter rules, short-lived credentials, audit logs, alerts, and MCP/HTTP proxies.
Apache-2.0 ★ 1 -
Audit-grade flight recorder for AI agents: policy on every MCP tool call, a signed hash-chained log, and OSCAL evidence
pypi MIT -
🔌 Exposes Hash-chained HMAC-signed audit log over MCP — for A2A (agent-to-agent) calls. Every tool-call, agent-handoff, decision gets a tamper-evident signed record. EU AI Act Art 12 automatic logs, DORA Art 17 ICT incident logs, ISO 42001 clause 9 monitoring — auditor-ready end-of-day attestations. By MEOK AI Labs.
pypi MIT -
Authorization, delegation and audit for AI agent actions over MCP: a policy on every tool call, short-lived agent tokens, human approvals and a tamper-evident audit log.
npm Apache-2.0 ★ 2 -
Work in progress: a hosted flashcards MCP server for teams (Python SDK 2.x). Built on mcp-flashcards-cloud, it adds owner/editor/viewer roles checked on every tool call and an append-only audit log.
MIT -
Uptime monitoring: monitors, outages, on-call schedules, status pages, maintenance and logs.
npm MIT -
A MCP server that can be called from Javascript so AIs can monitor javascript logs
-
A Model Context Protocol server that lets Claude and other AI assistants look up customers, orders and stock and take real actions, with a role for every API key, hard business rules in code and a tamper-evident audit log of every call.
MIT -
Self-hosted, open-source platform for running governed AI agents inside your own infrastructure. One model gateway for local or frontier models, a policy check on every model and tool call, a hash-chained audit log, OpenFGA permissions, MCP, and air-gapped installs. Python, Apache-2.0.
Apache-2.0 -
Call the local DeepSeek Harness (dsh) as a native tool from Codex over MCP: dsh_task / dsh_status, real per-call audit log, Windows-first.
MIT -
xclaudeaiupdated this year
xCLAUDE — A receipt for your AI. Local audit log for every Claude tool call.
npm MIT -
tokenmarkupdated this year
Drop-in observability and cost tracking for LLM API calls. Wrap your provider client, log every call with cost/tokens/latency to JSONL, query via CLI or MCP.
npm MIT -
agentsgategithub.com/agentsgate/agentsgateupdated this year
Undo for AI agents: an MCP proxy that checkpoints every tool call before it runs, so a bad `DROP TABLE` or `rm -rf` rolls back in seconds. Risk scoring, human-in-the-loop approval, tamper-evident audit log, dashboard, and MCP servers for PostgreSQL/MySQL/SQLite. Local-first, MIT, TypeScript, Node 20+.
npm MIT -
Persistent shell MCP server for Claude Code. Spawn headless shell sessions that retain full state (env vars, cwd, aliases) across tool calls. Sentinel-based output capture returns complete command output with exit codes. Auto-detects best available shell (Git Bash, pwsh, cmd). All sessions logged to ./logs/. The agency layer of the Velle triad.
Apache-2.0 ★ 2 -
Self-hosted Bitbucket for AI agents, the gap Atlassian's Cloud-only Model Context Protocol ("MCP") server leaves open, plus a leaner Jira and Confluence at 60 to 78% fewer tokens than raw REST, same latency. 58 tools on FastMCP over stdio, or one Auth0 OAuth 2.1 connector URL for the whole org, every call audit-logged: person, AI client, tool
GPL-3.0 -
Read-only MCP server that lets AI diagnose your Python backend: Postgres, Celery, Redis and logs in one call.
MIT -
salesloft-mcp-serverupdated this year
MCP server for Salesloft — people, cadences, accounts, activities, email templates, and call logging via the Model Context Protocol
npm MIT -
ai-toolinggithub.com/LonnaEdmond/ai-toolingupdated this year
MCP server: 31 typed tool contracts over field-operations data — schema validation, freshness checks, per-call audit logging. Python + FastMCP.
-
depositiongithub.com/georgedagher/depositionupdated this year
Local-only decision log for AI agent sessions -- zero API calls, deterministic decision extraction.
MIT ★ 1 -
auditguard-mcpgithub.com/ree2raz/auditguard-mcpupdated this year
A 7-stage compliance pipeline that gates every LLM tool call through RBAC, PII detection, policy enforcement, and audit logging — before and after execution.
Apache-2.0 ★ 1 -
relays error logs from other MCP servers when an LLM call fails and the returned error message is unclear, helping the model handle errors more intelligently.
npm -
sigil-protocolgithub.com/sigil-eu/sigilupdated this year
SIGIL security layer for AI agent tool calls — scans MCP tool arguments for leaked secrets, enforces policies, and writes audit logs
pypi MIT -
Fast, AST-based code-navigation: shape, public API, deps & call graphs, hybrid semantic search, structural rewrite, and log squeezing. MCP server included.
pypi MIT -
Local evidence gatherer for AI coding agents: one MCP call reads Jira, screenshots (OCR), Confluence, logs and the repo, and returns a single bounded dossier.
MIT -
splitwise-mcpgithub.com/Mannan-Goyal/splitwise-mcpupdated this year
Splitwise MCP server built from scratch in TypeScript. Direct REST API calls, no SDK — 27 tools covering expenses, groups, friends, comments, and more. Streamable HTTP transport with structured logging and PII masking.
-
Local-first MCP server that detects potential PHI flows into LLM, logging, and analytics calls in source code.
npm MIT -
senscritique-mcpgithub.com/kabylesystem/senscritique-mcpupdated this year
An MCP server for SensCritique: 17 tools an AI can call, public reads with no account, writes once you log in. On npm.
MIT ★ 2 -
MCP server where every AI agent tool call is authenticated and authorized by a Keycloak-issued JWT: role-gated tools, clear Forbidden errors, audit log
MIT -
Local meeting recorder and daily work log for macOS. Transcribes calls, writes notes, answers questions about them, serves an MCP. Nothing leaves your Mac.
MIT ★ 12 -
Let AI agents build and manage your WordPress site safely: MCP tools with preview, change log and one-call revert.
GPL-2.0 ★ 2 -
Flight recorder for AI agents: append-only event log of tool calls, errors, handoffs, file changes and token costs, with a query API, dashboard, and Claude Code hooks.
MIT -
agent_recordergithub.com/EdytaKucharska/agent_recorderupdated this year
Know what your AI agent actually did. Agent Recorder logs every tool call across your MCP servers — locally, to SQLite, without capturing prompts or reasoning. Debug failures, audit runs, and stop guessing.
MIT ★ 9 -
@rmotgi/compile-mcpgithub.com/rmotgi1227/compileupdated this year
MCP server that compiles repeat LLM work out of the agent loop. Logs receipts, identifies codifiable patterns, and routes hot calls to deterministic functions or local LLMs.
npm MIT -
orangerail-mcpgithub.com/kimhyeongrae0/orangerailupdated this year
Serve an ontology registry as an MCP server: typed read tools, destructive writes staged for human approval, and an audit log of every call.
npm MIT -
Native Timberborn mod and C# MCP server for AI agents: structured game state, guarded building and colony actions, research, simulation control and an in-game call log. No required third-party mods.
MIT -
Audits AI agent logs for wasted tokens and returns a quantified savings report. Finds redundant calls, retry loops, and duplicate reasoning steps. Hosted MCP server.
MIT -
Policy firewall for AI agents and MCP tools. Intercepts tool calls, enforces allow/deny/ask rules, redacts secrets, validates arguments, and writes auditable logs before agents touch your filesystem, GitHub, browser, or APIs.
Apache-2.0 ★ 6 -
k8s-port-forward-mcpgithub.com/oijusti/k8s-port-forward-mcpupdated this year
A Model Context Protocol (MCP) server that provides tools for discovering Kubernetes services and running kubectl port-forward sessions (optionally with separate log windows). It is designed for MCP clients/LLMs that translate natural language into structured tool calls.
npm MIT -
Fully local, privacy-first autonomous AI agent for laptop management — LangGraph reasoning loop, Ollama local LLM, ChromaDB RAG, MCP server, SHA-256 chained audit log. Zero API keys. Zero cloud calls.
Apache-2.0 ★ 5 -
Framework-agnostic Java observability sidecar agent: chat with a live JVM via MCP/web UI (metrics, slow calls, errors, logs, thread dumps, method traces). Self-hosted, attachable (Arthas-style). Apache-2.0.
Apache-2.0 ★ 3 -
Alpha MCP server that exposes contemplative-agent's memory cycle (episode logging, distillation into knowledge, identity, skills, rules and constitutional amendments) as tools that other AI agents can call, using the Anthropic API.
MIT ★ 1 -
MCP server for iOS app testing — see inside your running app: network calls, WebSocket frames, logs, storage, screen layout, and API mocking. Works with Claude Code, Cursor, Codex, Windsurf.
npm MIT ★ 1 -
Give your AI agents a valet key, not your master key. Local credential broker for AI agents (MCP) on macOS: Touch ID per credential, proxy calls, audit log.
Apache-2.0 ★ 1 -
MCP server for your *running* macOS Terminal.app session — list tabs, read scrollback, manage tabs, execute commands. Hardened three-tier safety policy (safe / requires_approval / forbidden) + audit log + native confirmation dialog. Shell state persists across calls.
MIT -
Switch AI models without breaking your product: find every model your code calls and when it retires, turn real usage logs into tests, compare and pick replacement models. CLI, GitHub Action and MCP server.
pypi MIT -
Andrea---AI-Healthcare-Receptionistgithub.com/ShreyKamdar0503/Andrea---AI-Healthcare-Receptionistupdated this year
Andrea is a production-ready, multi-agent voice AI receptionist for a dental clinic, built using VAPI, GPT-4o-mini, n8n workflows, and an MCP server. It handles real-world phone conversations including appointment scheduling, intake, insurance checks, and call logging with low latency and high reliability.
-
Connect Claude Code, Claude Desktop, VS Code and Cursor to ContextGate (https://www.contextgate.ai) in one command. Every tool call through it is logged and checked against your policies: governance checks, PII redaction and secret blocking. Free.
npm MIT -
Let your AI use your API keys through stuntkey without ever holding them: guided one-command setup, brokered API calls, a self-attack test and an activity log. Reduces risk; not independently reviewed.
pypi MIT -
Let your AI use your API keys through oathkey without ever holding them: guided one-command setup, brokered API calls, a self-attack test and an activity log. Reduces risk; not independently reviewed.
pypi MIT -
First MCP server for the CINC Systems (HOA / community association management) REST API. 13 tools across 9 resource groups: associations, homeowners, ACC, vendors, aged balances, call logs, documents, flagged collections, post charges.
MIT ★ 1 -
Open-source privacy testing for AI agents. Python SDK, CLI and MCP tools to detect data leaks across tool calls, memory, inter-agent messages and logs, with redacted reports and CI gates.
pypi MIT ★ 11