Ronda is live!! 🎉

Every good session with a coding agent holds something: the fix for a strange bug, the command that finally worked, the design call you'd never reconstruct. Then it disappears into a hidden folder you'll never open again.

I run five agents on a normal day, each writing to its own folder in its own format, none of it searchable together. So I built the library.

- 18 agents, one place. Ronda reads the session files they already write, nothing to export, no plugin.
- Search that lands on the message. Full-text across every transcript, ~26 ms p95 on an 800 MiB library.
- Resume without leaving. One click runs the agent's own resume command in a terminal inside - Ronda, locally or over SSH.
- Intelligence. Where the hours go, where agents get stuck, which errors keep recurring - derived locally, no model call.
- Your agents can read your history. A bundled CLI and read-only MCP server let Claude Code or Codex check whether you've hit an error before.
- Private by construction. Rebuildable SQLite index on your machine. Nothing uploaded, agent files only read.

Fully open source for macOS, Windows and Linux. There's a live preview on the site, the real interface on sample data, so you can try it in fifteen seconds without downloading anything.

Ronda

Every coding-agent session, searchable and open source

Every coding-agent session, searchable and open source