thread-keeper
Local MCP server giving multiple AI coding CLIs shared memory and coordination.
Tested on Linux
Use it
Linux x86_64
You need first
- Runtimes:
python>=3.11
install threadkeeper[semantic] via uv tool install
uv tool install 'threadkeeper[semantic]'
Check that it works
which thread-keeper-setup && which tk-agent-status && tk-agent-status --help 2>&1 | head -n 100; thread-keeper-setup --help 2>&1The output should contain “usage: tk-agent-status”.
First thing to try
tk-agent-status 2>&1; echo ---; thread-keeper-setup --dry-run 2>&1Add it to your agent
claude mcp add thread-keeper -- $HOME/.local/share/uv/tools/threadkeeper/bin/python -m threadkeeper.serverJSON for Claude Desktop, Cursor and other clients
{
"mcpServers": {
"thread-keeper": {
"command": "sh",
"args": [
"-c",
"exec \"$HOME/.local/share/uv/tools/threadkeeper/bin/python\" -m threadkeeper.server"
]
}
}
}Add thread-keeper as an MCP server in your AI coding CLI; you can preview the setup first with thread-keeper-setup --dry-run.
What it does
- Share memory between agents across sessionsNot shown working
- Coordinate agents running in parallel with signalsNot shown working
- Improve the skill library with automatic loopsNot shown working
How far pickfit got
Linux x86_64
- Installs
- Connects
- 0 of 3 features work
- Works offline
macOS and Windows: the project says it runs there; pickfit hasn’t tried it.
How pickfit tested it
Each run starts from a clean machine and follows the steps above. How pickfit verifies
Linux x86_64
An MCP client called one of its tools
- Replayed from scratch
- Works offline
- Checked on 1 Nov 2026
- Version 0.22.0
An MCP client handshake over stdio listed 144 tools, and a call to brief returned a result.
Differences the test machine hit
- README: thread-keeper-setup registers MCP server and copies hooks
In the test: setup reported no CLI detected and hook sources missing for tk-brief.sh etc. when installed via uv tool — server still starts