Every agent, one record

Run 20+ coding agents in one workspace

Claude Code, Codex, Cursor CLI, Copilot, Gemini, OpenCode, Amp, Aider, Goose, Cline, Devin, and more — launched from ORG-2 or ingested from their own on-disk history. An agent vendor can never be the neutral record across its competitors. We can.

20+
supported coding agents
15+
session formats ingested & replayed
1
record underneath all of them

GUI and TUI launchers

Launch supported CLIs from the desktop app with their own configuration, or run the native harness. Sessions land in the same record either way.

Session sources, unified

Scan history, inspect subagents, and control each source without switching apps — one manager for every tool's sessions.

Retroactive backfill

Install ORG-2 and your past months of sessions appear: each CLI's on-disk history is parsed and imported, so the record starts full, not empty.

Cross-agent memory

Knowledge agents build about your codebase is shared across sessions and agents — try a new model without starting from scratch.

Neutral by design

Claude Code will never record what Codex did. Teams run three to five agents; only an independent layer can hold the union of their work.

Usage and cost, rolled up

Token spend and cost aggregate across every runtime on the device — one place to see what your agents actually cost.