pond — Claude Code MCP server for session search
pond archives agent sessions and exposes them to search over MCP. Use it to retrieve an earlier fix or resume a session in another supported client.
Open on GitHubWhat pond does
pond gathers agent sessions into an archive stored in a local directory or your S3 bucket. It indexes messages, tool calls, and results so agents can search the collection over MCP; SQL queries are supported too.
Sessions can be restored into supported clients. Run pond init to configure storage and connections, then pond sync to index sessions from enabled adapters. Setup registers the MCP server in Claude Code. Restart the client afterwards so its tools load. On Windows, the POSIX skill-install line does not work in PowerShell 5.1; use the PowerShell instructions instead.
What it helps with
- retrieve an earlier fix from agent history
- archive sessions from multiple agents in one place
- resume a past conversation in another client
- search agent sessions through MCP or SQL
Example prompts for the agent
- Search pond for how we fixed this issue before, then apply that approach here.
- Check pond to see where we left off yesterday and continue the work.
- Look through pond for an earlier session about this problem.
How to install pond
Install
The MCP registry has no launch command for it: see the repository README
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about pond
How do I install pond in Claude Code?
Install it with brew install tenequm/tap/pond, then run pond init and pond sync. Restart Claude Code afterwards so the MCP tools load.
What does pond do?
It archives agent sessions, indexes them, and lets agents search them through MCP or SQL. Sessions can also be restored in supported clients.
Where does pond store session archives?
In a local directory or your S3 bucket. The archive retains session records, including messages and tool results.
