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.
-
A multi-agent system (planner, executor, critic) with an MCP tool server, a rule-based guard, and an evaluation harness.
-
codex-harness-mcpgithub.com/chapzin/codex-harness-mcpupdated this year
Local Codex MCP harness: contracts, persistent RAG memory, raw traces, verification records, governance policy and PASS/FLAG/BLOCK audits, observability reports, harness profiles, eval runs, Meta-Harness-lite promotion records, natural-language harness specs, MCP resources/prompts, multi-client installer, and completion gates.
MIT ★ 9 -
mcp-bigquery-evalsgithub.com/Umarfarook1/mcp-bigquery-evalsupdated this year
Read-only BigQuery MCP server with mandatory dry-run cost caps, agent-friendly structured errors, and a Spider/BIRD-style NL-to-SQL eval harness.
pypi MIT ★ 1 -
Agentic reconciliation-break triage system — bounded LangGraph state machine, MCP trust-boundary tool split, human-approval interrupt, and a real eval harness.
★ 1 -
Domain-pluggable LLM agent runtime with tool permissions, human approval, MCP server, and a repeatable eval harness. Local models (llama.cpp) and hosted APIs.
MIT -
Evidence-first AI agent for production root-cause analysis: multi-agent harness, signed evidence ledger, walk-back from the failure point, hybrid RAG, verifier + critic, evals with ablations, web UI, CLI and MCP.
Apache-2.0 -
evals-mcp-servergithub.com/cyanheads/evals-mcp-serverno update data
Agent-driven eval authoring via a new authoring-loop archetype — draft → subagent-review → surgical-revise → submit, with server-enforced self-consistent graders; compiles to standard harnesses.
smithery -
Governance-classifier eval harness on the Claude Messages API, with calibrated LLM-as-judge and an MCP server
MIT -
kensa-mcpgithub.com/satyaborg/kensaupdated this year
MCP server for kensa, the open source agent evals harness
pypi MIT -
An LLM eval harness graded by computed ground truth, not an LLM judge. Worked example: poker, served to models as MCP tools — gpt-4o-mini goes from 26/75 to 75/75 with them.
MIT ★ 1 -
skillsgithub.com/sattva2020/skillsupdated this year
Agent skills and MCP servers for AI-assisted development: harness audit (context, tools, memory, evals, guardrails), Dokploy API MCP
★ 3 -
geocheckgithub.com/AR-BABER/geocheckupdated this year
AI-search brand-visibility (GEO) checker - CLI + MCP server with a reproducible eval harness
MIT ★ 1 -
restaurant-agentgithub.com/illustraton916/restaurant-agentupdated this year
Permission-aware MCP server + spec-driven process + eval harness. Role-gated tools with structured refusals, prompt-injection evals against a real model — TypeScript
MIT -
mcp-judgegithub.com/jddavenportOpen/mcp-judgeupdated this year
A calibrated multi-persona LLM-as-judge, exposed as an MCP server and shipped with an eval harness that proves the calibration.
MIT -
The-Gate-VPSgithub.com/EditorialOS/The-Gate-VPSupdated this year
Production deployment layer for The Gate. Multi-tenant MCP server exposing the eval engine as agent-callable tools — scored eval harness, draft→eval→revise agent example, PRD, runbook, Docker.
NOASSERTION -
Permission-aware MCP server and agent harness with role-scoped tools, whole-plan validation, deterministic safety evals, and a React playground.
MIT -
copilot-studio-simulatorgithub.com/butelo/copilot-studio-simulatorupdated this year
Unofficial local Copilot Studio simulator: declarative agents (topics, tools, MCP, code execution, knowledge, child agents) + evals + auto-tuning, built on the pi harness SDK
MIT ★ 1 -
MCP server for Indian bank IFSC data — lookup, offline validation, branch search, bank directory — plus a routing eval harness that scores its own tool descriptions.
MIT -
Self-hosted research assistant: hybrid RAG (BM25 + vectors) with verifiable citations, grounded AI briefs, eval harness, MCP server. Runs free with Groq + local embeddings.
MIT -
Context-APIgithub.com/EditorialOS/Context-APIupdated this year
Memory and brand-context layer for AI agent systems. Multi-tenant MCP server — 39 tools auto-generated from a live OpenAPI spec. PostgreSQL, per-workspace isolation, scored eval harness.
NOASSERTION -
oncall-copilotgithub.com/sasi-nemani/oncall-copilotupdated this year
Provider-agnostic AI on-call assistant — RAG + a tool-using agent + an MCP server + an eval harness — runs on Claude, OpenAI or OpenRouter. A hands-on learning project.
GPL-3.0 ★ 1 -
Multi-agent workflow engine — LangGraph router, domain agents, MCP tools, human-in-the-loop gates for financial actions, compensating sagas and an eval harness, built with FastAPI and Python.
MIT -
Branch SQLite like git — copy-on-write forks, checkpoint/rollback/promote, crash-tested capture. Stock SQLite files on local disk or S3. Built for AI agents and eval harnesses.
go Apache-2.0 ★ 6 -
EDGAR MCP: six tools over SEC filing data with an eval harness
MIT -
AI agent for portfolio managers: answers questions about a fund vs its benchmark and risk limits, and writes a daily morning brief. Built on a custom MCP server, with a hand-written agent loop and an eval harness.
MIT -
ukvet — AI Due-Diligence Agent for UK Companies. Cited risk reports from Companies House data, with an MCP server and eval harness.
pypi MIT -
An autonomous LLM fund that scores its own stock predictions — Brier score, calibration curves, eval + ablation harness — and serves its full decision history over MCP. LangGraph pipeline with deterministic risk guardrails.
★ 1 -
Text-to-SQL agent over public Hedera mainnet data. Ask in English, get the SQL back as the citation — with an eval harness whose cases are proven to reject careless queries. Also an MCP server.
MIT ★ 1 -
mcp-rag-pgvector-demogithub.com/fr3kchy/mcp-rag-pgvector-demoupdated this year
Production RAG pipeline: pgvector + nomic-embed-text + Claude API with MMR reranking and eval harness
★ 1 -
evalbenchgithub.com/cognis-digital/evalbenchupdated this year
Offline LLM / agent eval harness with regression gates
NOASSERTION -
Self-hosted temporal knowledge-graph memory for AI coding agents — Graphiti/Neo4j, MCP, local embeddings, retrieval eval harness
NOASSERTION -
Codebase Q&A with (path:line) citations — the capstone of a from-scratch AI-engineering series. Built in 8 eval-gated stages (chat → RAG → agents → hardening → production), provider-agnostic and offline-first, with 4 extensions (MCP, agent harness, context management, local models).
MIT ★ 1 -
canopygithub.com/kruslim/canopyupdated this year
An MCP server exposing vehicle-diagnostic / CAN-bus domain logic as agent tools, with a LangGraph orchestration layer and a calibrated human-in-the-loop eval harness. Built above a source-agnostic seam so OBD swaps to raw CAN with no changes above it.
★ 1 -
Local multi-agent pipeline (Ollama) that turns an idea into a validated tech spec. MCP server, structured outputs, eval harness.
MIT -
Agentic RAG over French bank annual reports: LangGraph agent + MCP server + hybrid search, with page-level citations and an eval harness.
-
Gauntletgithub.com/Deep-De-coder/Gauntletupdated this year
Adversarial eval harness for any LLM agent pipeline — Claude, OpenAI, or your own. CLI + REST API + MCP server for Cursor/Antigravity.
NOASSERTION ★ 1 -
Question answering over your own documents: FastAPI, pgvector hybrid search with RRF, cited answers, LangGraph agent, MCP server and an eval harness
MIT -
p-layersupdated this year
Memory for AI agents with rules that are kept, not just written. P0-P6 layered rules enforced in code. SQLite + FTS5 + pluggable embeddings, PostgreSQL backend, MCP server, eval harness. (GitHub: p-layer)
pypi MIT -
Production-grade context engine for AI agents — hybrid RAG, reranking, token-aware assembly, MCP server, eval harness
-
MCP server for legal intake triage: deterministic tools, conflicts gate, eval harness
Apache-2.0 -
RAG over docs with verified citations, hybrid BM25 + embedding retrieval, an eval harness, and an MCP server. Python, FastAPI, Gemini/Groq.
NOASSERTION -
gq-insight-mcpgithub.com/yusufdxb/gq-insight-mcpupdated this year
MCP server for semantic search + grounded, citation-faithful answers over customer-research interviews, with a built-in eval harness. Built for Great Question.
MIT -
Reproducible RAG & Agent harness: 932-chunk corpus, 30-question frozen eval, 17s zero-API regression gates, self-hosted retrieval/generation/tracing.
-
Natural-language analytics over a governed Snowflake warehouse: LangGraph agent, MCP tool server, role-based PII masking, eval harness, Terraform on AWS.
-
Autonomous game playtesting agent: plays a game, hunts bugs with generic oracles + LLM, writes reproducible reports. Includes seeded-bug eval harness and MCP server.
MIT -
Autonomous research & experiment engineering agent platform: LangGraph multi-agent orchestration, 3 custom MCP servers, sandboxed execution, HITL guardrails, ResearchOpsBench eval harness
MIT -
cligithub.com/prjct-app/cliactive
The agentic harness for AI coding agents — intelligence is rented, the harness is owned. Runs the full work cycle: intent briefs, bounded RAG context from real code intelligence, persistent memory, preventive guardrails, and performance evals. One CLI + MCP server, SQLite inside. Claude Code, Codex, Gemini CLI, Cursor & OpenCode.
npm MIT ★ 1 -
Answers RFPs from your own product docs with cited, quality-checked drafts. Go API, Python workers, pgvector, Redis Streams, MCP server, LLM eval harness.
MIT ★ 1 -
tearsheetgithub.com/Wynelson94/tearsheetupdated this year
Local web-to-markdown MCP server for Claude Code — scrape, crawl, map, search, extract. Token-efficient, guard-railed against silent omission, trust-qualified by a falsifiable eval harness.
MIT ★ 1 -
posthog-mcp-minigithub.com/josephruocco/posthog-mcp-miniupdated this year
An MCP server that serves PostHog's JS/Web SDK docs as agent-optimized context, with an eval harness that measures it against a naive baseline.
MIT -
Multi-tenant RAG & MCP server on .NET: pgvector search, cited answers, MCP tools for Claude Code, every retrieval change measured with an eval harness
★ 2 -
Multi-tenant B2B call analytics platform — LangGraph pipeline, MCP server, pgvector RAG, JWT auth with Postgres Row-Level Security, and automated LLM-as-judge eval harness
-
Double-entry ledger with Kafka fraud detection and a local-LLM investigation agent (LangGraph + MCP), with an eval harness
MIT -
A company brain for AI agents, with proof the output is good. MCP server: shared memory, hybrid search, email loop, and an eval harness.
Apache-2.0 -
inbox-mcpgithub.com/jairo21991-debug/inbox-mcpupdated this year
MCP server exposing a CRM-style inbox (contact search, conversation retrieval, BM25 full-text search) to LLM agents — with a retrieval eval harness (Hit@k/Recall@k/MRR).
MIT -
Deployed retrieval-augmented agent: ADK + authored MCP server + pgvector on Cloud Run, with an eval harness
MIT -
Local-first RAG assistant: answers only from your own notes/docs/code, every claim cited, graded by an eval harness. OpenAI-compatible gateway (free local by default), exposed over MCP.
MIT ★ 6 -
matterpilotgithub.com/yashpatil582/matterpilotupdated this year
Legal AI matter platform: pluggable workflow packs (court-notice intake, contract review), matter-scoped pgvector RAG, Office add-ins, MCP server, eval harness
MIT -
Local OpenAI-compatible embeddings and reranking on Apple Silicon: bge-m3 hybrid search, Qwen3, a retrieval eval harness and MCP memory for agents.
MIT ★ 1 -
semantic-code-searchgithub.com/n1ra7/semantic-code-searchupdated this year
Self-hosted, zero-cost semantic code search + RAG with an MCP server — FastEmbed embeddings, Qdrant vector DB, incremental indexing, and eval harness. Python.
MIT