Skip to content

Launch FAQ

karurikwao edited this page Jul 6, 2026 · 1 revision

Launch FAQ

What is RunWitness?

RunWitness is a local-first witness layer for autonomous agent work. It records observable actions and exports receipts.

Is RunWitness an agent?

Not primarily. Agents do the work; RunWitness records policy decisions, commands, file changes, approvals, timelines, and receipts.

Is RunWitness a sandbox?

It has sandbox foundations and opt-in container execution. It is not a universal OS sandbox for every run path.

Is RunWitness upstream-official for OpenClaw, Hermes, Codex, or Claude?

No. The integrations are official from the RunWitness repo side. They are not upstream-official until those upstream projects accept, list, or ship them from their own channels.

What should early users try?

  • run a local witnessed command
  • inspect the generated receipt
  • try a policy check
  • validate the OpenClaw plugin package
  • connect the MCP server to an MCP-capable host
  • give feedback on missing receipt evidence

Clone this wiki locally