Home

RunWitness Wiki

RunWitness is a local-first witness layer for autonomous agent work: agents can run tasks, while RunWitness records policy decisions, commands, file changes, approvals, tests, timelines, and receipts.

Use this wiki for longer launch and adoption material. Keep the README short; use these pages for operational detail.

Pages

• Getting Started
• Core Concepts
• Integrations
• Receipts and Policy
• Security Model
• Launch FAQ
• Upstream Official Path

Current Status

RunWitness is ready for a GitHub alpha launch as a local-first tool. It is not a hosted SaaS control plane, universal OS sandbox, universal secret boundary, or complete nested-agent tracer.

The launch promise is simple: run a task, observe every action RunWitness can see, approve risky steps, and produce a receipt.