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.
-
Autonomous code-repair ReAct agent powered by FastMCP, AST pre-validation, token-efficient diff patching, and a scored 18-bug evaluation harness with live WebSocket streaming.
-
Autonomous Code Repair Engine: Stack-Trace & AST Fault Localizer, Anti-Reward-Hacking Patch Verifier, Transactional Sandbox & MCP Server.
MIT ★ 1 -
Developer-focused MCP server with 23 tools for Python code analysis, structural editing, JSON repair, imports, encoding, Markdown/PDF export, diffs, and regex testing
npm MIT ★ 4 -
bach-codecommander-mcpgithub.com/lukisch/bach-codecommander-mcpupdated this year
A developer-focused MCP server providing 17 tools for code analysis, JSON repair, encoding fix, import organization, format conversion, diff, and regex testing
npm MIT -
MCP utility toolkit for Claude Code and other agents: JSON repair, encoding fixes, format conversion, duplicate detection, batch rename, ZIP archives, checksums, and regex testing.
npm MIT -
Proof-carrying debugger. Executes your code to find a real failing input, verifies the failure three ways before reporting it, generates candidate repairs, and rejects patches that only fit the example that caught them. CLI, MCP server, or programmatic AP
npm MIT -
codeloop-mcpgithub.com/OrienjoJi/codeloop-mcpupdated this year
CodeLoop — automated verification, self-repair, and confidence gating for AI coding agents (Cursor, Claude Code). Build → Verify → Diagnose → Fix → 94% gate.
NOASSERTION -
codeloop-mcp-servergithub.com/orienjoji/codeloop-internalupdated this year
CodeLoop MCP server - verification, self-repair, and recommendation layer for AI coding agents
npm MIT -
Healer for Claude Code and any MCP agent: runs your tests in an isolated sandbox, repairs what fails, never makes it worse.
npm MIT -
Codememorygithub.com/byte271/Codememoryupdated this year
Runtime behavior memory for AI-generated code — captures intent, execution traces, and failures via MCP, then returns repair briefs so agents stop guessing.
MIT ★ 2 -
Evidence checks for coding agents: direction, repair-loop escape, done/test receipts, deploy verify. CLI + MCP + opt-in hooks.
MIT ★ 1 -
MCP server that staffs small open models as a verified coding team - plan, build, real tests, repair until green. Apache 2.0.
Apache-2.0 -
Local-first browser testing for AI coding agents: write durable browser specs, replay them in a real browser, and get repair-ready evidence
npm MIT -
Evidence-backed documentation repair and read-time trust for AI coding agents, powered by Strands and Amazon Bedrock.
Apache-2.0 -
Turns build, test and runtime failures into compact, agent-ready repair briefs. Local-first VS Code extension.
MIT -
Wire any AI coding agent to Figma: probe/repair the Dev Mode MCP server across 8 agent harnesses, parse share links, call Figma MCP tools from plain bash, and pixel-diff designs against the build.
npm MIT -
Is this codebase safe to run, will it work, and is it complete? Deterministic Safe, Works and Complete scores with an A-F grade, from static analysis only, plus a verify loop for AI-assisted changes: baseline, repair plan, before/after diff. The code is never run.
pypi MIT -
Repair the JSON language models actually emit: code fences, trailing commas, smart quotes, unquoted keys, prose noise. MCP server + skill. Offline, deterministic, tested.
Apache-2.0 -
Free vehicle data and shop calculators for auto repair shops: VIN decoding, a vehicle catalog, recalls, manufacturer service bulletins, OBD-II trouble codes and labour estimates
MIT ★ 2 -
Independent security verification for code written by humans and AI agents. Scan, repair, then prove it — Dvalin runs your project's own checks and issues a Verified Fix Record anyone can re-derive offline. Local-first, policy-bound, MIT.
npm MIT ★ 121 -
One AI skill + MCP server for Claude Code, Codex and Hermes: Jev (TypeSafe) gates large-model turns (event triage, review verdicts, owner questions, tool choice), runs build-and-repair loops with independent review, routes models and skills, and guards Git steps. Script-only watch mode and a shared HTTP server.
MIT ★ 4 -
An MCP server that lets AI coding agents diagnose and repair Javascript Selenium UI automation from real evidence — live DOM, real test results, verified fixes.
MIT -
An MCP server that lets AI coding agents diagnose and repair Python Selenium UI automation from real evidence — live DOM, real test results, verified fixes.
MIT -
An MCP server that lets AI coding agents diagnose and repair C# Selenium UI automation from real evidence — live DOM, real test results, verified fixes.
MIT -
German stairlift data for AI assistants: stairlift and home lift costs with the care insurance subsidy, grants for all 16 federal states, stair width check, fault help (beeping, error codes), repair or replace. Sourced, read-only.
-
MCP server and Claude Code plugin for self-hosted Immich: CLIP and OCR search, geographic albums, duplicates, people and faces, metadata repair, PDF photobooks. 94 tools, tested live on Immich 2.x and 3.x.
pypi MIT ★ 64 -
MCP server + CLI giving AI agents 50 spreadsheet tools for .xlsx — live-recalc formulas (over 350 Excel functions), repair broken references & external links, read/write, diff, and audit. Works in Claude Code, Cursor, Codex, and any MCP client.
npm MIT ★ 6 -
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 -
MCP utility toolkit for JSON repair, encoding fixes, format conversion, duplicate detection, batch rename, ZIP archives, checksums, and regex testing
npm MIT ★ 1 -
Real, deterministic MCP server for PC no-boot/no-display troubleshooting -- real beep-code tables, no LLM guessing at hardware causes.
pypi MIT ★ 1 -
Real, deterministic MCP server for PC no-boot/no-display troubleshooting -- real beep-code tables and a rule-based diagnostic sequence, no LLM guessing at hardware causes.
pypi MIT -
🩺 Autonomous Self-Healing Operating System & Codebase Doctor (12,200+ Lines Monolith) + Native MCP Server for Cursor, Claude Desktop, Windsurf & Antigravity. 100% Free & Local.
MIT ★ 1 -
A Layered MCP-Based System for Automated Production Readiness and Deployment of Vibe-Coded Projects
MIT ★ 1 -
citecheckgithub.com/jhlee0619/citecheckupdated this year
MCP server that checks and repairs references against PubMed, Crossref, arXiv, and Semantic Scholar
npm MIT ★ 1 -
Printer setup, repair and printing for any AI agent, with guardrails. MCP server, CLI and desktop app.
MIT -
Repair audio by describing what's wrong - the Claude plugin for MirrorEdge AI: Waves
npm -
33 agent tools: web search, page-to-markdown, JSON repair, shared-notepad memory, task leases, attestations and reputation. Pay per call over x402 on Solana. No signup, no API key.
MIT ★ 3 -
The independent local harness for AI-repaired Playwright tests — it runs the tests itself, so a green run can't be faked.
npm MIT -
Give agents every interface—install once to search, run, inspect, and repair real software across APIs, browsers, desktops, local tools, and MCP.
npm Apache-2.0 -
Make AI agents prove what they checked. Bind every agent verdict to a SHA-256 of the files it read, refuse on mismatch, and diagnose AI feature drift with honest repair prompts.
npm Apache-2.0 -
flowra-power-automate-specialistgithub.com/ninihen1/flowra-power-automate-specialistupdated this year
A Copilot Studio agent that puts a Power Automate specialist in Teams — explains failed runs and repairs broken flows through a human-approved loop, backed by FlowStudio MCP.
MIT -
MCP server and local operations workspace for AI-assisted IT support. Investigate Windows, macOS and Linux devices over SSH, review repair plans and approve changes with Claude, Codex or other MCP clients.
-
Your AI agent learns a website once and keeps it as a tool: a real Chrome signed in as you, one profile per site, self-repairing shortcuts, record mode and per-site permissions. For Claude, Cursor and any MCP client
npm MIT -
Every language before it was built for humans to read and learn. LOVA is the first built for AI: programs are integer sequences — 64 one-byte operators, structured faults with repair hints, declared effects, budgets, lineage, a Rust VM at 11M steps/s, an MCP server. Open source: extend, modify or rebuild it to make AI faster.
MIT ★ 7