Chatback
FreeNot checkedA local-only MCP server for Cursor that indexes agent transcripts and surfaces unfinished or on-hold chats, providing tools to search, review, and pick up where
About
A local-only MCP server for Cursor that indexes agent transcripts and surfaces unfinished or on-hold chats, providing tools to search, review, and pick up where you left off.
README
Find unfinished and on-hold Cursor agent chats — then pick them up again.
Local-only MCP server for the Cursor community. Indexes your agent transcripts under ~/.cursor/projects, scores open loops (aborted, errored, interrupted, open todos, hold language), and exposes tools so the agent can show you a Chatback inbox.
No cloud. No upload. MIT licensed.
Why this exists
Generic chat search helps you find old threads. chatback helps you finish them — ranked by what looks unfinished or explicitly on hold.
Install (30 seconds)
Cursor Marketplace / plugin
Once listed, install Chatback from Marketplace. Until then, add to ~/.cursor/mcp.json:
{
"mcpServers": {
"chatback": {
"command": "npx",
"args": ["-y", "github:princemuichkine/chatback"]
}
}
}
Restart Cursor (or Developer: Reload Window).
Local development
git clone https://github.com/princemuichkine/chatback.git
cd chatback
npm install && npm run build
Point MCP at your build:
{
"mcpServers": {
"chatback": {
"command": "node",
"args": ["/absolute/path/to/chatback/dist/index.js"]
}
}
}
Or symlink as a Cursor plugin:
ln -s "$(pwd)" ~/.cursor/plugins/local/chatback
MCP tools
| Tool | Purpose |
|---|---|
list_open_loops |
Ranked inbox of unfinished / on-hold chats |
search_chats |
Keyword search across indexed transcripts |
get_chat |
Full transcript body by chat id |
recent_chats |
Recently updated chats |
mark_loop |
Mark hold, done, or ignore (local only) |
reindex |
Rescan transcript folders |
Resource: chatback://inbox — markdown summary of top open loops.
CLI (debug without MCP)
npx -y github:princemuichkine/chatback --help
npx -y github:princemuichkine/chatback --print-inbox 10
Open-loop signals
- Last agent paragraph — cleaned user-facing closing (monologue and
[REDACTED]stripped); inbox shows up to ~480 chars - Agent invitation closings on that paragraph (
want me to,if you want,say the word,when you're ready, …) — not generic “follow-up migration” notes turn_endedwithabortedorerror- Last message is from the user (interrupted mid-turn)
- Open
TodoWriteitems with non-empty content - Strong hold phrases in user text (
on hold,come back later,close this chat, …) - Your local
holdmark viamark_loop
Recent chats rank higher; threads older than ~30 days are gently demoted.
Marks are stored in ~/.cursor/chatback/marks.json. The search index lives at ~/.cursor/chatback/index.db (schema v3; atomic writes; per-file index errors are collected, not thrown).
Privacy
Reads only Cursor agent transcript files on your machine. Never sends data off-device.
Publish checklist
- cursor.directory — community listing
- Cursor Marketplace — submit open-source repo
License
MIT — see LICENSE.
Installing Chatback
This server has no published package — it is built from source. Open the repository and follow its README.
▸ github.com/princemuichkine/chatbackFAQ
Is Chatback MCP free?
Yes, Chatback MCP is free — one-click install via Unyly at no cost.
Does Chatback need an API key?
No, Chatback runs without API keys or environment variables.
Is Chatback hosted or self-hosted?
Self-hosted: the server runs locally on your machine via the install command above.
How do I install Chatback in Claude Desktop, Claude Code or Cursor?
Open Chatback on unyly.org, pick your client tab (Claude Desktop, Claude Code, Cursor) and press Install — the config is generated automatically, no JSON editing.
Related MCPs
Fetch
Web content fetching and conversion for efficient LLM usage.
AWS KB Retrieval
Retrieval from AWS Knowledge Base using Bedrock Agent Runtime.
by modelcontextprotocolSpring AI MCP Server
Provides auto-configuration for setting up an MCP server in Spring Boot applications.
llm-analysis-assistant
A very streamlined mcp client that supports calling and monitoring stdio/sse/streamableHttp, and can also view request responses through the /logs page. It also
by xuzexin-hzCompare Chatback with
Not sure what to pick?
Find your stack in 60 seconds
Author?
Embed badge for your README
Browse similar
All ai MCPs
