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.
-
MCP Server generated by mcp.ag2.ai
★ 1 -
Model Context Protocol server for Open Policy Agent (OPA). Author, validate, debug, and explain Rego policies through any MCP-compatible client.
npm MIT ★ 8 -
Open-source approval gateway for AI agent tool calls: policy, human approval, one-time execution grants, and audit evidence.
npm Apache-2.0 ★ 1 -
@checkoutmd/walletgithub.com/gitlevo/checkoutmdupdated this year
The credential wallet for the open agent ecosystem. MCP-native, local-first, policy-driven.
npm Apache-2.0 -
AgentGuard is an open-source, local-first MCP security gateway that protects AI agents with policy-based tool permissions, human approvals, audit logs, and policy simulation. Developed by Belfeki.tn.
Apache-2.0 -
No shell. Only actions you defined. Open source operations gateway for AI agents: signed, typed actions over MCP, per-host policy checks, batched execution and a hash-chained audit trail.
MIT -
Policy-enforcing MCP server that lets AI coding agents safely read repos, apply patches, and open pull requests across Forgejo, Gitea, and GitLab.
Apache-2.0 -
The open authority layer + SDKs for AI-agent payments. Thin Python/TS clients, an MCP server, framework adapters (LangChain/CrewAI/Hermes/OpenClaw/…), and @sardis/reference — a pure policy simulator + AP2/TAP/x402 verifiers showing exactly how Sardis decides if an agent may spend. The hosted engine that moves money is private.
npm MIT ★ 6 -
Self-hosted on-call alerting and incident escalation. Open-source PagerDuty and Opsgenie alternative with on-call schedules, escalation policies, SMS, voice and push notifications, a mobile app and an MCP server for AI agents. Built on GoAlert.
-
Neutral, cross-vendor control plane for AI agents that spend money — one policy, one sub-second freeze, one tamper-evident audit ledger across every agent-wallet backend. Open-core: SDK + MCP + x402/AP2.
npm NOASSERTION ★ 1 -
wp-oss-AI-agentgithub.com/homgorn/wp-oss-AI-agentupdated this year
Open-source AI agent operating system for WordPress — capability-first, policy-aware, self-hosted. 121 tests, 31 components, R1 complete. Replace paid WordPress.com MCP with free self-hosted AI automation for content, SEO, WooCommerce, plugins, and DevOps.
MIT ★ 1 -
An MCP server that lets AI agents operate a multi-cluster Kubernetes fleet through an Open Cluster Management hub, with policy, approval, and audit between the model and your clusters.
Apache-2.0 ★ 17 -
Open-source multi-platform ads management for AI agents — connect your ad accounts, manage them via MCP + CLI with policy-guarded writes
npm Apache-2.0 ★ 19 -
Open-source agent runtime — SSH-native isolation, eBPF egress policy, Kubernetes + LXC backends, GPU passthrough, MCP-native CLI
pypi Apache-2.0 ★ 318 -
An open mail client for people and agents. Native Mac app, IMAP push, Apple Mail import, and an MCP server with a send policy and approval step.
MIT -
MCP servers that let LLM agents operate Juju safely: server-side policy guardrails, audit log, Prometheus-driven autoscaling, and an eval harness for open-weight models.
Apache-2.0 -
An open, user-owned memory protocol for every app and agent, with provenance, bitemporal truth, policy-first retrieval, and safe fallback.
npm Apache-2.0 -
agentic-vaultgithub.com/AliProgrammin/agentic-vaultupdated this year
Open-source MCP server that lets AI agents call APIs without ever seeing the credentials. Policy-enforced secret injection into HTTP calls and subprocesses, with encrypted audit trail.
AGPL-3.0 ★ 1 -
safe-solana-mcpgithub.com/liander-ai/safe-solana-mcpupdated this year
An MCP server giving AI agents safe, policy-gated access to Solana (with a Chain Fusion read on ICP): reads are open, transfers are capped, allowlisted, simulated, and returned unsigned. No keys, no broadcast.
MIT -
Open-source AI security platform providing perimeter defense for LLMs and AI agents through swarm analysis, policy enforcement, adversarial testing, and real-time threat detection.
pypi AGPL-3.0 ★ 21 -
An open source agentic wallet for Solana. Claude and other AI agents pay USDC through it without ever holding the private key: the vault keeps the key, checks each payment against the owner's policy, and signs.
MIT -
🧠 Open-source context engine that feeds AI agents with policy-compliant enterprise knowledge. MCP-native, GDPR & EU AI Act ready, self-hosted.
Apache-2.0 ★ 5 -
Precomputed answers for event streams, kept current in a SQLite file as events arrive: an open-source policy language with SQL and Go runtimes, a log reducer, usage metering, AI agent traces and an MCP server.
Apache-2.0 -
Micro ECF is the open local policy and context harness for coding agents: dependency-free CLI plus MCP server, durable ECF.md, source maps, budgets, approvals, and Agent OS exports. No cloud or wallet.
Apache-2.0 -
Open-source self-hosted context-governance runtime for coding agents: local MCP plus CLI, cited context maps, retrieval and eval artifacts, policy boundaries, and optional Triptych OS (Agent OS) preview exports. No hosted runtime or wallet authority.
Apache-2.0 ★ 1 -
Open-source AI agent security gate: deterministic ALLOW/BLOCK/ESCALATE authorization before every tool call, plain-English policy compilation, tamper-evident audit ledger. MCP security gateway + Claude Code/Cursor guardrails. Rust, Apache-2.0.
Apache-2.0 -
The agentic meta-harness — freeze the model, evolve the harness. An open runtime that routes each query to the cost-optimal model, evolves its own harness (planner/context/reviewer/retry/tool/memory/score policy) and autonomously repairs code, then orches
npm MIT -
HTTP was built for browsers. YEA is built for agents. YEA (Your Explicit Approval) is an open protocol for AI agents acting on behalf of people: intents, proposals, signed policy, human consent, undo, token budgets. Works with Claude Code, Cursor, Codex and any MCP client.
Apache-2.0 -
Extracts claims from SOPs, policies and regulations, has a person approve or reject each one, then serves the approved set to agents over MCP - each claim carrying a byte range to its source sentence. Open Knowledge Format (OKF) bundles, no database.
Apache-2.0 ★ 18 -
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 -
Open-source MCP server that lets your AI agent shop for you while you keep control of the money. The agent fills the cart and reaches payment, then hands you a UPI QR to approve on your phone. Your PIN never touches the AI. Local-only, with a signed spending policy.
pypi Apache-2.0 -
The open routing layer for AI tools. MCP server that routes agent tool calls (search, extract, browse) across 8 providers per policy — cost, latency, pinning, deny lists — with failover and the routing decision on every result. One Go binary, BYOK, no account. Source-available, BUSL 1.1 → Apache 2.0.
npm NOASSERTION ★ 2 -
Open source input firewall for AI agents, beta. A local scanner checks text, code, PDFs, images, QR codes, audio and video with 1,565 patterns across 117 categories and reports findings and incomplete scans. A Claude Code hook blocks secret material in tool calls and the paths and hosts your policy lists before tools run.
pypi MIT ★ 11 -
Limetry is open-source agent action governance: check tool intents against policy before irreversible runs. Returns allow, deny, or approval_required with signed receipts and privacy-safe audit. TypeScript SDK, CLI, and MCP for IDE like Cursor, VS Code, GitHub Copilot, Claude. Optional CI, SQL, and Shopify adapters. MIT — self-host your own node.
npm MIT ★ 1 -
Read-only MCP guardrail for Kubernetes and Terraform. It returns allow, require approval, or deny. Python and OPA share one policy, and this process never applies a change.
MIT -
Attack. Defend. Verify. Red-team your AI agents, enforce zero-trust policy at runtime, and prove each fix holds in CI. Includes an MCP proxy.
NOASSERTION -
Proposal-only payment-intent governance for AI agents: deterministic policy, demo human review, hash-linked audit evidence, REST, MCP, and mock outcomes—without a live payment rail.
MIT -
team-task-managergithub.com/anan5093/team-task-managerupdated this year
A production-ready full-stack project and task management system with React, Express, MongoDB, and a LangGraph-orchestrated Python AI Swarm service.
MIT ★ 1 -
Open-source authority layer for AI agents: who may approve what, how decisions escalate, and the evidence.
Apache-2.0 -
Broker-neutral process, evidence, and control layer for financial agents: Capital Constitutions, a hash-chained decision ledger, offline-verifiable proofs, and a stdio MCP companion. Research prototype. Not a trading bot. Not affiliated with Robinhood.
Apache-2.0 -
Deterministic UI policy, discovery and quality gates for coding agents.
npm MIT ★ 1 -
Open-source local-first MCP policy gateway for controlled ChatGPT access to local files and terminal sessions.
Apache-2.0 ★ 5 -
The Ostrom server, in Kotlin: gates, signed receipts, policies, the MCP relay and the API.
Apache-2.0 -
Reference implementation: one MCP agent shared by many tenants, where every tool call gets a 30-second PostgreSQL credential, handed over as a single-use OpenBao wrapping token and limited to one tenant by row-level security. Keycloak, SPIRE, Agentgateway, OPA, OpenBao and kagent on a local kind cluster.
-
Local-first co-browse browser for AI agents. Same real Chromium tabs for you, MCP, and Playwright — policies, human takeover, trajectory export.
MIT -
Turn the world's AI laws into checks your code, agents and CI can run. Open, cited obligations for 17 regimes, the AI_GOVERNANCE.md standard, a CLI, MCP server, GitHub Action, agent rails and deadline calendar. Not legal advice.
Apache-2.0 ★ 2 -
Open-source MCP server and CLI for ManyChat, with a fail-closed Meta messaging-policy guard.
npm AGPL-3.0 ★ 1 -
One MCP server for the databases and REST APIs you already have (Postgres, MySQL, SQLite, MongoDB, OpenAPI), with per-table policy, read-only enforcement, redaction and audit.
npm MIT ★ 1 -
Local-first self-healing QA for web apps: Electron desktop, CLI, and MCP with isolated Git worktrees, SQLite evidence, and policy-aware GitHub repair.
AGPL-3.0 ★ 4 -
The reliability layer for what AI agents do: exactly-once execution, action policies, human approval and a full audit trail. Self-hosted, MCP-native, free Community edition (MIT/Apache-2.0).
pypi NOASSERTION ★ 3 -
🔌 Exposes Open source license compliance over MCP — SPDX parsing, compatibility checking, attribution generation, policy enforcement. MIT.
pypi MIT -
sideguardgithub.com/alisaitteke/sideguardupdated this year
MCP guard & vibe coding security tool for Cursor and Claude Code. Fail-closed human-in-the-loop approval for shell commands and MCP tools—local daemon, YAML policy, no cloud.
MIT ★ 2 -
Voice-first J.A.R.V.I.S. desk companion for the Waveshare ESP32-S3 Touch AMOLED 1.75C — Gemini Live, reactive round UI, physical privacy, motion, OTA, and policy-gated MCP tools.
Apache-2.0 ★ 9 -
Your company's AI rules in every AI tool: one reviewed policy file served to Claude, ChatGPT and Codex while plans are made, plus open security skills and Claude Code hooks. No telemetry.
MIT ★ 2