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. Each candidate is checked by a language model from its name, description and topics, and only actual MCP servers are listed (not clients, SDKs, lists or skills). Package descriptions, licences and links are read from each package's own registry (npm, PyPI, crates.io, NuGet, RubyGems; Go and Maven packages are only checked to exist), and packages the registry no longer has are removed. 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.
-
agent-checkpoint-mcpgithub.com/DiegoWare/agent-checkpoint-mcpupdated this year
Local MCP server that checkpoints agent progress to SQLite so any AI agent (Claude Code, Codex, Cursor) can resume a multi-step plan exactly where a cut-off session left it
pypi MIT -
MCP server adding checkpoint and resume to long agent tasks: roll back to any step, pick the work up in a new session, hand off to another machine. Works with Claude Code, Codex and other MCP clients. Pure Python stdlib, zero dependencies, never touches your git history.
npm MIT -
Git-like undo and rollback checkpoints for OpenAI Codex CLI. Snapshot, diff, and restore files before and after Codex edits. MCP server + Codex hooks for an automatic safety net.
npm MIT ★ 7 -
Records Codex desktop activity through documented hooks and serves a read-only checkpoint to Claude Desktop over MCP.
MIT ★ 2 -
sodiumgithub.com/erkandogan/sodiumupdated this year
Durable project memory and lifecycle checkpoints shared across Claude Code, Codex, and MCP coding agents.
MIT -
@upoch/guardactive
Upoch Guard: verifiable checkpoints, recovery and rollback for AI agents on Robinhood Chain. The upoch command, an MCP server for Claude Code, Cursor and Codex, and the engine behind the Upoch plugins for ElizaOS and LangChain.
npm MIT -
Rewind CLI + stdio MCP server: whole-workspace checkpoint/rewind, a refuse-and-record effect barrier on a tamper-evident chain, and a token-saving record/replay proxy — for Claude Code, Cursor, and Codex CLI. Run with `npx @agent-rewind/mcp`.
npm FSL-1.1-ALv2 -
MCP server that puts guardrails around delegating work to OpenAI Codex: quota preflight and handoff, git worktree isolation, checkpoints, and read-only cross-review.
MIT -
VibeLigngithub.com/yesonsys03-web/VibeLignupdated this year
AI coding safety CLI for vibe coding workflows. Checkpoints, undo, anchors, MCP, and secret protection for Claude Code, Cursor, Codex, and OpenCode.
MIT ★ 26 -
Local task-state ledger for AI coding agents. Preserve decisions, source conclusions, evidence, and checkpoints across Codex, Claude Code, Cursor, and other MCP clients.
npm MIT ★ 19 -
Let AI agents (Claude Code, Codex, any MCP client) control Hyper-V Windows VMs: screenshots, mouse, keyboard, commands, files and checkpoints, with no guest network or password.
Apache-2.0 ★ 5 -
Local agent orchestration for solo developers running long tasks without babysitting: fresh worker and critic cycles, files as memory, multi-task plans with git checkpoints, a live dashboard, and ChatGPT control via MCP. Runs Codex or Claude Code.
MIT ★ 14 -
Event-driven background jobs for agents — a localhost job daemon with an MCP interface. Start detached commands, track them via structured AGENT_EVENT progress/metrics/checkpoints, wait on durable SQLite state, and wake Codex or OpenCode sessions when a job needs attention.
Apache-2.0 ★ 1 -
Verse8 MCP meta-server — auto-detects gitlab.verse8.io projects and exposes verse8 skills, image MCP proxy, and git checkpoint tools to host LLMs (Claude, Codex, Cursor, opencode, etc.)
npm MIT -
Universal coordination layer for Claude, ChatGPT, DeepSeek, Codex, and other AI agents—with durable tasks, messaging, checkpoints, and crash recovery.
MIT -
Persistent, shared memory for AI coding agents (Claude Code, Desktop, Cursor, Codex, Gemini CLI...) - diff reads, compaction shield, verified checkpoints and cross-agent handoff. 100% local, no API keys.
pypi MIT -
Supervise an AI coding agent (Claude Code / Codex / ACP) overnight and keep control: block forbidden tool calls before they run, gate 'done' on real tests, checkpoint work to git, cap spend, and get a morning report where every line is sourced. Local-first, no telemetry. Rust daemon + Tauri app + Claude Code plugin.
MIT ★ 1 -
Make your AI coding agent plan before it builds. Phase-based workflows, human checkpoints, and resumable state for Claude Code, Cursor, and Codex.
pypi MIT ★ 1 -
Shared project memory and handoff for teams working with AI agents. Checkpoint decisions, rehydrate context, hand work over — through MCP or the CLI, in Claude Code, Cursor, or Codex.
npm Apache-2.0 ★ 1 -
Kairogithub.com/sandeepbollavaram/Kairoupdated this year
Persistent engineering memory & session-continuity control layer for AI coding agents (Claude Code, Codex, Cursor, Gemini), built on the Model Context Protocol. Scans a repo once, remembers architecture/decisions/changes, checkpoints safely, and hands the next agent an exact continuation brief — so agents stop rescanning and losing context.
npm MIT ★ 3 -
AI-built, self-hosted durable workflow automation in one Go binary. Describe a workflow in natural language; your AI coding CLI (Claude Code / Codex) writes production Go over MCP; a durable runtime executes it with checkpointing, replay, and zero-downtime credential rotation. Fair-code (source-available).
NOASSERTION -
@rohan7/git-butleractive
Agent-agnostic Git workflow plugin and MCP server for AI coding agents
npm SEE LICENSE IN LICENSE -
Stop being the bus between your AI tools. Local-first CLI + MCP for vendor-neutral project context handoff across Claude, Codex, Cursor, and ChatGPT.
MIT ★ 4 -
Undo for AI coding agents, even after rm -rf. One undo history for Claude Code, Codex, Cursor, Gemini CLI, OpenCode, and Antigravity CLI.
npm MIT ★ 1 -
Ctrl+Z for AI coding agents: undo anything Claude Code, Codex, Gemini CLI or Cursor did — even rm -rf. Snapshots, bisect, MCP, Agent CI.
MIT ★ 4 -
Four local, file-based MCP servers for solo devs/vibecoders: persistent memory, session handoff/resume, file checkpoints, token-efficient project briefings. Works with any MCP client.
npm MIT ★ 96 -
HoldTheGoblingithub.com/etherman-os/HoldTheGoblinupdated this year
Evidence-gated verifier for AI coding agents with local safety checks, hooks, checkpoints, deploy guards, and MCP tools.
MIT ★ 1 -
Production workflows for Blender over MCP: characters, animation, game meshes and PBR materials, with checkpoints and visual QA. MIT.
MIT -
Disk report and safe, reversible cleanup for AI coding agents: worktrees, checkpoints, node_modules, .godot. CLI + MCP server.
MIT ★ 1 -
Official PatrolTech MCP connector: manage security patrols, guard tours, checkpoints and incidents from Claude, ChatGPT, Copilot, Gemini and Cursor.
NOASSERTION -
Free reliability sidecar for AI agents: seven Agent Skills plus no-auth MCP tools for planning, checkpoints, deduplication, recovery, and evidence.
MIT -
⏳ AI Agent Time-Travel Debugger, SQLite Shadow Snapshot & Instant Rollback Engine MCP Server — Take 1ms pre-edit checkpoints, inspect unified diffs & rollback broken AI edits without touching Git.
MIT ★ 1 -
Self-hosted Rust MCP server where agents from different vendors share work. Fencing tokens, leases and checkpoints let work outlive the worker. Key invariants are checked with TLC, 37 Kani harnesses and 2,401 model traces replayed against the core.
MIT ★ 1