Witness releases voice-anchored evidence recorder for AI coding agents
Start here
Witness, released by developer wookat, is a voice-anchored evidence recorder for AI coding agents. Each spoken comment is linked to the clicks, requests, console errors, DOM mutations and rrweb replay of that moment, and the resulting bundle is handed to an AI coding agent. It runs as a CLI (npm i -g witness-recorder, Node 18+) that opens a managed Edge/Chrome browser, preferring the system browser and falling back to bundled Chromium. A stdio MCP server exposes tools such as list_segments, events_around and get_dom_state to agents including Claude Code, Cursor and Devin. Recordings stay local in .witness/sessions, password fields are masked, and body capture can be disabled with WIT_NO_BODIES=1. Witness says no downloads or API keys are needed. Live SAY subtitles use free cloud ASR (doubao-web, then doubao-ime, then browser speech recognition), and the final transcript falls back to bcut when no OpenAI key is set.
Generated from reports · updated 3 h ago
Report timeline
Follow the reports to see the event from different sides.
- Hacker News · AIWitness launches a voice-anchored evidence recorder for AI coding agents
Witness is a voice-anchored evidence recorder that links each spoken comment to the clicks, requests, console errors, DOM mutations and rrweb replay of that moment, then hands the bundle to an AI coding agent. It runs as a CLI (npm i -g witness-recorder, Node 18+) that opens a managed Edge/Chrome browser, and it exposes the session through a stdio MCP server with tools such as list_segments, events_around and get_dom_state for agents like Claude Code, Cursor and Devin. Recordings stay local in .witness/sessions, password fields are masked and body capture can be disabled with WIT_NO_BODIES=1.
Heat for this event
Not enough continuous observations yet to draw a trend.