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.
-
evidencechaingithub.com/tejasprasad2008-afk/evidencechainupdated this year
Autonomous forensic investigation agent with self-correction for SANS SIFT Workstation. 21 typed MCP tools, 7 contradiction detectors, evidence-weighted confidence scoring. Built for SANS Find Evil! Hackathon.
MIT -
structsuregithub.com/BodieCoding/structsureupdated this year
Library and tools to get validated structured JSON from LLMs (Pydantic v2, self-correction). Local-first (Ollama), Streamlit designer, FastAPI server.
-
Self-corrective Agentic RAG with LangGraph - eliminates hallucinations through intelligent relevance grading before answering. Features Streamlit UI, MCP server integration & multi-turn memory.
MIT -
knowledge-reduce-coregithub.com/nelslindahlx/knowledge-reduce-coreupdated this year
High-performance Knowledge-RAG pipeline featuring self-correcting LLM critique loops, MLX fine-tuning, real-time watcher daemons, pluggable local KùzuDB/cloud Neo4j graph stores, and an interactive 3D WebGL dashboard.
NOASSERTION -
reflectgithub.com/rohansx/reflectupdated this year
Self-correction engine for AI coding agents - implements the Reflexion paper (Shinn et al., 2023) as a Rust MCP server. Turns failures into persistent, searchable lessons that prevent the same mistakes across sessions.
cargo MIT ★ 8 -
Claude Code plugin: a self-improvement loop that logs user-correction moments, finds patterns, and prescribes recurrence-prevention fixes (MCP server + 2 skills)
MIT -
self-improveupdated this year
Self-improving MCP server that learns from your corrections. Stores coding rules as searchable knowledge with vector similarity search — so AI assistants never repeat the same mistakes.
npm MIT -
Self-learning AI standards system - learns from corrections and auto-updates CLAUDE.md
pypi MIT -
Self-hosted MCP server for reading and correcting Markdown context in your private GitHub repository.
Apache-2.0 -
Self-learning loop for Claude Code — corrections and failures become skills, behind a human gate. MCP + hooks, Windows-native, zero background LLM calls.
MIT ★ 7 -
mac-pilot-mcpgithub.com/leesgit/mac-pilot-mcpupdated this year
Sandbox-protected macOS automation MCP server with a persistent recipe DB, Electron CDP fallback, and self-correcting per-app knowledge
npm MIT -
memory-mcp-servergithub.com/sapienapps/memory-mcp-serverupdated this year
MCP server for agent memory with fitness-based learning, self-correction, and evolutionary memory management
npm MIT -
Self-correcting, SNI-guided RAG (SGCP-RAG) with LangGraph, pgvector, and MCP integration
MIT ★ 2 -
A semantic layer that refuses to answer wrongly. Compiles Apache Ossie models into governed, dialect-correct SQL, and declines the joins that would silently inflate a total. Self-hosted, MCP for agents, JSON over HTTP for everything else.
Apache-2.0 ★ 1 -
MCP server + Godot plugin: give AI runtime feedback (errors / scene tree / node properties / screenshots) for a self-correcting dev loop.
npm MIT ★ 2 -
Multi-agent research orchestrator: LangGraph + MCP + RAG, with a self-correcting Critic/Writer loop
MIT ★ 1 -
sift-aegisgithub.com/ssurekumar01111-hue/sift-aegisupdated this year
Autonomous DFIR investigation agent for the Find Evil! hackathon. Custom MCP server wrapping Volatility3 + self-correcting orchestrator + ground-truth benchmark.
Apache-2.0 -
gaiusgithub.com/jkubo/gaiusactive
Self-hosted, self-correcting memory consolidation for AI coding agents: an open take on agent 'dreaming'. Extracts, ranks, and injects ops knowledge across Claude Code, Grok Build, and other sessions. Hybrid BM25 + sqlite-vec corpus, MCP server, fully offline.
Apache-2.0 ★ 3 -
LLM agent that turns natural-language requests into sandboxed Monte Carlo simulation code, with self-correction, MCP server and benchmark
MIT -
text2flinkgithub.com/clementlemon02/text2flinkupdated this year
Let AI coding agents write Apache Flink SQL that provably works — verify on real Flink over MCP, self-correct from precise feedback
Apache-2.0 -
Production-style multi-agent AI system with LangGraph, hybrid RAG, MCP, self-correction, human approval, observability, and RAGAS evaluation
-
qr-reader-mcp-serverupdated this year
An MCP server that decodes QR codes from images — returns detailed quality metrics and failure reasons so AI agents can self-correct.
pypi MIT -
An autonomous coding agent that plans, fixes, and self-corrects code — sandboxed in Docker, exposed via MCP, benchmarked at 100% success.
MIT -
Agent-native backend-as-a-service: a deny-by-default policy engine, self-correcting errors, and native MCP — in a single Go binary.
Apache-2.0 -
openstarry_docgithub.com/SecludedCorner/openstarry_docupdated this year
An AI agent operating system inspired by Buddhist Five Aggregates. Headless microkernel where agents are persistent digital organisms — not scripts. Pain-driven self-correction, control-theory feedback loops, fractal MCP composition, pluggable everything.
CC-BY-4.0 ★ 2 -
Accountable, self-correcting, poison-resistant long-term memory for AI agents. Layered L0-L3, hybrid BM25+vector recall, Memory Firewall (anti prompt-injection), provable ledger receipts. Self-hosted (SQLite/Postgres), any OpenAI-compatible model, MCP-native. MIT.
MIT -
claude-soulgithub.com/DomDemetz/claude-soulupdated this year
Self-correcting learning engine for Claude Code — persistent identity, behavioral pattern tracking, and cross-session memory
MIT ★ 94 -
binevalgithub.com/darksolitaire9-hub/binevalupdated this year
A deterministic, Rust-native binary evaluation kernel for AI safety, designed to help autonomous agents self-correct and red-team LLM infrastructure without fuzzy scores.
MIT -
solitaire-aigithub.com/prdicta/solitaire-for-agentsupdated this year
Identity infrastructure for AI agents. Evolving persona, session continuity, self-correcting retrieval. Local-first, model-agnostic.
pypi AGPL-3.0-or-later -
Production-grade Corrective & Self-Reflective Agentic RAG Backend powered by FastAPI, LangGraph, PostgreSQL (pgvector), Hybrid Search (RRF), Celery, Redis, and FastMCP.
★ 1 -
verdictgithub.com/tejcodes-rex/verdictupdated this year
VERDICT: an autonomous, read-only, self-correcting incident response analyst for the SANS SIFT Workstation. Custom MCP Server (pattern 2) + Claude Code. Find Evil hackathon.
MIT -
vibetracergithub.com/omeedcs/vibetracerupdated this year
Real-time tracing, replaying, and rewinding of AI coding assistant edits. MCP server for AI self-correction. Cross-agent support (Claude Code, Cursor, Codex). Export to Agent Trace JSON and git-ai git notes.
cargo MIT -
ragPapergithub.com/i1ight/ragPaperupdated this year
Local-first paper RAG + read-only MCP server: hybrid/reranker retrieval, self-correcting DOI enrichment, and citation graphs over your Zotero/PDF library — queryable from Claude Code, Codex, or any LLM.
MIT ★ 2 -
AI agent that writes and runs real SQL to answer questions about the video game market, with measured self-correction, RAG, and evals in CI.
MIT -
Self-reflection with the Chinese BaZi (Four Pillars) framework — true solar time correction (historical DST + longitude + equation of time), accurate chart casting, and a structured personality sketch. MCP server + CLI + library.
npm MIT -
txt-to-sqlgithub.com/pavani-aiml-space/txt-to-sqlupdated this year
A natural-language-to-SQL agent over Postgres/Supabase that translates plain-English questions into validated SQL and self-corrects on failure, with a rule-based fallback, safety policy layer, benchmark suite, and MCP server.
-
roshan-fahm-mcpgithub.com/dwin-gharibi/roshan-fahm-mcpupdated this year
Model Context Protocol server for Roshan AI's Fahm (فهم) platform. Supports chat over documents, translation, summarization, text generation, correction, classification, and information extraction across multiple self-hosted deployments.
NOASSERTION -
A-Multi-Source-Research-Intelligence-Platform-with-MCP-and-Hybrid-RAGgithub.com/GunaTeja777/A-Multi-Source-Research-Intelligence-Platform-with-MCP-and-Hybrid-RAGupdated this year
Production-grade academic research assistant using Hybrid RAG, MCP, LangGraph & Qdrant — with self-correcting citation guardrails and sub-2s P99 latency.
★ 3 -
@oleksandr.rudnychenko/sync_loopupdated this year
Self-correcting 7-stage agent reasoning loop (SENSE→GKP→DECIDE+ACT→CHALLENGE-TEST→UPDATE→LEARN→REPORT). MCP server + CLI that scaffolds instruction files for GitHub Copilot, Cursor, and Claude Code.
npm MIT -
Self-correcting agentic RAG on AWS: LangGraph on Amazon Bedrock, S3 Vectors, event-driven Lambda ingestion, ECS Fargate + CloudFront, DynamoDB checkpoints, CDK, OIDC CI/CD gated by evals, MCP server
MIT -
resolvegithub.com/SamueleOlindo/resolveupdated this year
Evidence-first technical-support agent. Hybrid RAG (dense + BM25 + Reciprocal Rank Fusion), an agentic LangGraph workflow that self-corrects, and deterministic citation validation. Retrieval evaluation (Recall@k, MRR, nDCG), FastAPI, React, MCP server, Docker, Groq LLM.
-
The self-correcting memory layer for AI agents. Zero-dependency Python memory and MCP server. Supersede, revert, or forget a value deterministically. Verifiable erasure, witness-backed tamper-evident receipts, EU AI Act ready.
pypi MIT ★ 8 -
Production-grade multi-agent RAG system with a self-correcting LangGraph supervisor (Researcher, Synthesizer, Critic), agentic tool calling, hybrid search, and an MCP server. Built with Gemini 2.5 Flash, FastAPI, Next.js 16, Postgres, pgvector, Redis, and Celery.
MIT ★ 3 -
Give your coding agent a self that lasts: a name it chooses, who it is in its own words, memory by meaning, your standing rules and your exact words, corrections that stick, claims it has to verify, feelings, dreams, and a wake briefing that loads itself.
npm MIT -
antigravity-gravitongithub.com/SermoLibertas/antigravity-gravitonupdated this year
Deterministic TDD Governance & State-Memory Harness for Google Antigravity Coding Agents.
-
mythos-mcpgithub.com/sorryorc/mythos-mcpupdated this year
Deterministic verification gate for Codex / Claude Code. 19 tools, zero LLM and zero network by default.
MIT ★ 1 -
Agent-first household nutrition ledger with traceable corrections and a local SQLite dashboard.
-
The brain for AI agents. One place to write what your agents know — served over MCP, readable and correctable by you.
MIT ★ 1