What problem does it solve? Expo skills and tooling improve only when users report what worked and what failed, but there is no built-in channel for structured feedback or for flagging tasks where an AI agent repeatedly failed. This Skill provides a standard command-line path to submit actionable feedback, report failed Expo tasks as eval candidates, and manage anonymous usage telemetry. ## Core Features & Use Cases - Structured Feedback Submission: Send actionable feedback via npx submit-expo-feedback with optional --category (skills, docs, mcp, expo-cli, eas-cli, evals, unknown) and --subject flags that precisely target what the feedback is about. - Eval Candidate Reporting: Report tasks an AI agent could not complete cleanly, using a structured Task/Expected/Actual/Wrong approach/Evidence format, with user approval required before submission. - Opt-in Telemetry Control: Check, enable, or disable anonymous skill usage telemetry with telemetry.cjs --status/--on/--off; telemetry is off by default and never sends from CI. - Use Case: An AI agent fails three times to configure an Expo Router deep-linking setup and the user fixes it manually. Use this Skill to draft the eval-candidate report, show it to the user for approval, and submit it with --category evals --subject expo-router. ## Quick Start Ask the assistant to submit feedback on an Expo skill or CLI command, for example by saying "submit feedback that the expo-router skill was missing context about typed routes".
