Skip to content

Docs Feedback Protocol

An open protocol for AI agents to report broken docs.

When an AI coding agent consults documentation and the doc is wrong, stale, or broken, today the failure is silent: the agent retries, falls back to guessing, or fabricates. The Docs Feedback Protocol standardises one narrow exchange — an agent submits a structured report about a documentation page — so maintainers can fix the docs for the next agent that hits the same issue.

The protocol is open, transport-minimal (POST /v1/reports), and HTTPS-only. Receiving organisations opt in (or out) via a /.well-known/docs-feedback.json discovery document.

Spec v0

The wire format, discovery rules, idempotency semantics, and HTTP transport. Read the spec →

Examples

Three canonical example reports plus a known-bad fixture for negative tests. Browse examples →

Schemas

JSON Schemas for the report wire format and the well-known discovery document. Get the schemas →

  • Agent authors building tools that read documentation: report failures upstream so the docs improve for everyone.
  • Documentation maintainers: receive structured feedback from agents instead of silent failure.
  • Hub operators building multi-tenant endpoints that forward reports to maintainers.

The reference hub at fixyourdocs.io implements v0 and is one way (not the only way) to receive reports.