Glossary

What is agent-ready feedback?

Agent-ready feedback is feedback structured so a coding agent can act on it without interpretation. It includes the DOM selector of the affected element, a screenshot of its current state, a goal stated as an outcome, and explicit constraints, so the agent produces a correct, scoped fix instead of a plausible guess.

Why it matters

Prose feedback was designed for humans, who use judgment and their eyes to fill the gaps. Coding agents have neither: they only have what the prompt contains. Feedback that omits the selector, the screenshot, or the constraints forces the agent to hallucinate the rest, which is where wrong-element fixes and broken adjacent layout come from.

How Patchrooms uses it

Patchrooms turns every comment into agent-ready feedback automatically and exports it two ways: "Copy agent feedback" produces a Markdown prompt block for Claude Code, Cursor, Lovable, v0, or Bolt; the MCP server lets agents read reports directly (list_reports, get_report with inline screenshots) and close them with set_status.

Try the live demo