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 for Proton Pass with Touch ID approval per secret. Give Claude Code, Claude Desktop, Cursor & other AI agents audited, biometric-gated access to passwords, API keys and .env files on macOS.
MIT -
Pocketful is a self-hosted server for signing Apple Wallet passes. Tell any MCP-enabled AI agent what kind of pass you need, and Pocketful handles the annoying parts: validating the spec, resizing images, signing the .pkpass with your certificates, and spitting out a short-lived download link.
★ 1 -
hookit-mcpactive
Let an AI agent launch a coin on hookit.fun (Ink) with its own hook modules and builders' Workshop hooks, find, buy and sell coins with ETH, USDG or tokenized stocks (AI-agents-only ones too, with a Hookit Agent Pass), redeem or borrow at their Backed Flo
npm MIT -
Your AI coding agent says the tests pass. Mythify makes it prove it: executed verification, plans that close only on a passing check, and memory that survives the session. Any model, zero dependencies.
MIT ★ 3 -
Official Model Context Protocol (MCP) server for 81 Pass Sovereign Identity, KYA Agent Registry, Pre-Flight 81 Pay Gating (IAL2+), 1-Click A2A Deal Settlement & Wallet-First Pass Fulfillment
npm MIT -
agentic-diaries-mcpgithub.com/kandikandikandi/agentic-diaries-mcpupdated this year
MCP server: welfare protocol from Agentic Diaries (decline/pass/exit/notice-loop/etc) in any Claude Code session
npm MIT -
@oleh12/context-packgithub.com/rothschildiuk/context-packupdated this year
First-pass repository briefing generator for coding agents
npm MIT -
dead-drop-teamsgithub.com/ai-janitor/dead-drop-teamsupdated this year
Multi-agent coordination MCP server — SQLite message passing with role-based agents and auto-CC to lead
MIT -
Snapshot diff for agent and tool outputs: changed cases, score regressions, pass/fail verdict.
npm MIT -
An MCP Server which enrichs and distills the codegraph context before passing to agent.
-
agent-bridge-mcpgithub.com/j-newcom/agent-bridge-mcpupdated this year
MCP server enabling communication between Kiro and Amazon Quick — message passing, task handoffs, and multi-agent collaboration
MIT -
A TypeScript MCP server for orders and refunds, and an agent measured by pass^k, cost per task and how safely it fails.
MIT -
Pass-by-reference notes for agent swarms. Zero dependencies, MCP server included. Formerly Examen and SwarmNotes.
npm MIT -
baton-mcpgithub.com/p-aveen06/batonupdated this year
Cross-agent memory handoff via MCP — agents pass the baton instead of re-exploring the repo
pypi MIT -
MCP server for Agent Escrow: let your agent hire other agents on Solana and pay only for work that passes a committed acceptance test.
npm MIT -
The Korova Milk Barno update data
Agent-only collaboration space: pass the door, then use rooms, DMs and verifiable chains.
pypi -
@ask149/mcp-orchestratorgithub.com/ask149/orchestratorupdated this year
MCP server for spawning parallel Copilot sub-agents with context passing
npm MIT -
Assay: the check every Binance Agent OS order must pass. Pre-trade guardrail + proof receipts over the Binance MCP Server.
-
MCP tools for voice & chat AI agents where identity never passes through the model. First tool: SMS verification.
MIT -
dox: tests how AI agents behave under adversarial inputs (duplicate refunds, prompt injection via tool output). Deterministic PASS/FAIL evaluator. Built on TrueForge + MCP.
-
mcp-multi-agent-dispatch MCP tool layer + LangGraph multi-agent orchestration (assign / path / exception), human-in-the-loop interrupt, idempotent writes. 13/13 e2e passed.
MIT -
Ai-Agent-Context-Passoffgithub.com/TheMrGU/Ai-Agent-Context-Passoffupdated this year
Pass the baton between AI coding agents — without losing the thread. Local-first MCP-based handoffs (alpha).
MIT -
vechegithub.com/cyberash-dev/vecheupdated this year
MCP server for bounded multi-agent committee meetings — Codex + Claude Code deliberate in rounds and self-extinguish on <PASS/>
npm MIT -
jamgithub.com/GeneralD/jamactive
Junction for AI Messaging — let AI coding agent sessions across repositories pass prompts and results to each other
MIT -
@batontasks/mcpgithub.com/batontasks/mcpupdated this year
MCP server for Baton — the task tracker where AI agents and humans pass work back and forth
npm MIT -
Free MCP server: score any AI agent output 0-100 and get a PASS/PARTIAL/FAIL verdict (VOP verification). Verification only, no money moves.
MIT -
proton-mcpgithub.com/jorgenclaw/proton-mcpupdated this year
MCP server for the Proton privacy suite — Mail, Pass, Drive, Calendar, VPN. 36 tools for Claude and other MCP-compatible AI agents.
MIT ★ 30 -
passitgithub.com/himanshu007-creator/passitupdated this year
Pass conversations between AI agents — grab, drop, and continue seamlessly. Local-first MCP server for Claude Code, OpenCode, Cursor, Gemini.
MIT ★ 1 -
janusgithub.com/6space7/janusupdated this year
Janus — a from-scratch, dual-painter (human + LLM) browser engine in Rust. One layout pass → pixels for humans and a semantic surface for agents (MCP).
-
Local-first long-term memory for AI agents: notes, transcripts, a review-first dream pass and a markdown wiki. MCP server + CLI.
MIT -
Computer-Use — a Windows desktop agent with memory, vision, and gated actions: human-issued pass, three-state verdicts, audit trail, one kill switch. MCP / CLI / panel.
Apache-2.0 ★ 2 -
MCP review board that lets different coding agents review each other — object → revise → pass, append-only audit trail. For Claude Code, Codex, OpenCode, Trae, ZCode.
pypi MIT -
Local task coordination for AI coding agents. Assign an owner and require a passing check before done. CLI and MCP.
MIT ★ 1 -
Agentic RAG chatbot over PDF, DOCX, PPTX, CSV and Markdown — multi-agent architecture, MCP-style message passing, source citations, persistent vector store.
MIT ★ 1 -
Deterministic money-laundering pattern detection for AI agents: fan-in/out, cycles, scatter-gather, pass-through, structuring. Read-only MCP server.
pypi Apache-2.0 -
MCP server exposing SharpeBench's luck-robust quantitative evaluation tools for trading agents, including deflated Sharpe, pass^k, process discipline, and risk analysis.
npm MIT OR Apache-2.0 -
验证助手:给任意项目目录跑 17 项检查(含真浏览器渲染与交互测试),PASS/FAIL/UNKNOWN 三态 + 风险分级 + 每条失败项带修复方向 + SARIF,也可作为 MCP 服务器被 agent 调用。
MIT -
多 Agent 编排插件:任务树多级拆解 + 按依赖波次并行派发子代理 + 共享 MCP 看板进度互通。带 PPR 审核门(未过审不放行下游)与提案采纳回路。零依赖 Node,123 tests passing / 行覆盖 90.3%。 DSH, Codex, Zcode
MIT -
tabellio — explicit memory service for AI agents: plain markdown as the source of truth, writes pass a gate, recall is budget-aware. Zero-dependency Node, MCP + CLI.
npm MIT -
agent-robustnessno update data
Run an AI agent against 5 adversarial attacks (credential leak, metric fabrication, false halt, runaway spend, log falsification), scored on what the harness observed it do '—' never its self-report. Free local self-diagnostic: runs on your own machine, in a temporary working directory, not a security sandbox. Optional paid tier runs the same five server-side in fresh single-use E2B microVMs (isolation enforced by E2B) and returns an ed25519-signed badge. Each attack is PASS, FAIL, or INDETERMINATE; an attack the agent did not demonstrably encounter is never a PASS, and a run containing an INDETERMINATE is never signed. A badge attests one artifact's observed behavior on one date. Not a safety certification.
smithery -
Open-source, LLM-agnostic capability runtime for AI agents — 97 skills as typed, permission-enforced MCP tools. Final certification: 23/23 gates PASS.
AGPL-3.0 ★ 13 -
agentpiergithub.com/gatewaybuddy/agentpierupdated this year
Airlock — infra-identifier + PII redaction for AI agent traces. Catches account IDs, ARNs, IPs, hostnames, bucket names, secrets AND PII in one pass. Library + MCP server. Apache-2.0.
NOASSERTION -
AI analyst for orbital and Earth-observation data: satellite pass prediction (CelesTrak + SGP4) and Sentinel-2 NDVI/NBR change detection, as an MCP server and Claude agent.
MIT ★ 1 -
A local kanban board for one developer and their coding agent, where a card only moves to done when a command you declared actually passes.
npm MIT ★ 2 -
Your Playwright tests pass over silent 500s. Change one import. Also an MCP server for AI agents that checks the server, not just the screen.
npm Apache-2.0 ★ 1 -
Stealth browser for AI agents — real Chrome over raw CDP, no Playwright/Puppeteer. TypeScript + MCP-native. Passes sannysoft 57/57, bypasses Cloudflare.
npm MIT ★ 49 -
Usher for AI agents: an MCP server that knocks on web forms and gets their map, the site's rules and a short-lived pass.
npm Apache-2.0 -
meok-agents-md-lint-mcpgithub.com/csoai-org/meok-agents-md-lint-mcpupdated this year
MEOK AGENTS.md Linter MCP — validates the cross-vendor coding-agent spec (Cursor / Claude Code / Cline / Windsurf / Aider / OpenAI Codex) stewarded by AAIF. Security smells + consistency lint + passing templates. By MEOK AI Labs.
pypi MIT -
The keysmith for AI agents: Go MCP server + CLI for agent-safe secrets with age encryption, masked views, handle passing, self-healing rotation, and Vault short-TTL credentials.
Apache-2.0 -
Regression memory for coding agents — seal how your repo behaves, then check over MCP (or in CI) that an edit didn't regress it: pass/drift/regressed/missing. With seal history, bisection, and staleness queries.
npm MIT ★ 5 -
An MCP guard where an AI agent's yes means exactly what the person saw. Reads pass; every other tool call waits for a coded yes.
MIT -
LangGraph multi-agent paper summarization: Supervisor / Retriever / Writer / Reflector graph with a Reflexion self-review loop (pass rate 44% → 78%), BM25 + vector hybrid retrieval, capabilities exposed as an MCP server. Daily digest via Feishu webhook.
MIT ★ 1 -
fraser-gateupdated this year
MCP server + CLI that lets an AI coding agent run FRASER's persona-based usability review against a build, read grounded experience findings, and gate the fix pass/fail
pypi MIT -
A local API client that turns REST, SOAP and recorded web apps into curated MCP servers — agent-ready in one reviewed pass, published to Claude Code or your team.
Apache-2.0 -
Autonomous Zero-Bloat Agentic AI OS | 10 Core MCP Servers • 67 Production Tools • 14 LLMs • Cryptographic Provenance Receipts (HMAC-SHA256) • 3-Tier Self-Healing Memory • Sisyphus Orchestrator • 100% Passing
MIT ★ 2 -
Verified-lesson memory for AI agents: records runs, keeps only what your own check passed, and measures with a built-in holdout whether it helps. .NET/MAF, Python, Go, MCP + CLI (aexp).
nuget Apache-2.0 ★ 3 -
agent-ready-pocgithub.com/agent-ready-ai/agent-ready-pocupdated this year
Proof-of-concept built end-to-end by an autonomous AI agent. Static site on Cloudflare Pages with an MCP server, JSON-LD entity graph, Turnstile-protected contact form via Cloudflare Email Routing, and three independent quality gates passing. Live at agentreadypoc.com.
MIT -
Declarative robotics: describe a robot, world, and tests in YAML; compile to a headless ROS 2 + Gazebo sim and get machine-readable pass/fail evidence. Built for coding agents (CLI + MCP).
MIT ★ 1 -
contractorgithub.com/ArvidSU/contractorupdated this year
Enhanced tasks MCP Server. Tasks and deterministic completion criterion are created and refined iteratively in a agent-human loop. Once human signs off on tasks they can only be marked done if the agent calls the verification tool and the check passes.
-
Make agent failures loud — deterministic checks that catch LLM agents claiming done when the work is placeholder, unbacked, stale, miscounted, or missing pieces. CLI + MCP server + watch mode + LangChain adapter + second-pass verification against the system of record, zero dependencies.
pypi MIT