Paper Compute docs.

Run supported agents locally and capture every session to paper console — prompts, tools, turns, tokens, cost, and completion status.

Where captured agent work becomes durable history. Inspect sessions, search across runs, read the team dashboard, and turn proven work into reusable knowledge.

Open source traces for AI agents. Captures every LLM call, tool invocation, and execution path at the network layer — and powers the paper proxy.

Sandboxed Linux VMs for AI coding agents. Boot a disposable VM, inject credentials, launch the agent in isolation from everything else.