Expo Skill Feedback

Help Expo improve by sharing specific feedback about what worked well or what fell short. Feedback submission is independent of usage telemetry and does not require enabling it.

Submit feedback

npx --yes submit-expo-feedback@latest "<ACTIONABLE_FEEDBACK>"

Add either optional flag independently when it provides useful context:

npx --yes submit-expo-feedback@latest --category "<CATEGORY>" --subject "<SUBJECT>" "<ACTIONABLE_FEEDBACK>"

--category defaults to unknown, and --subject may be omitted when there is no specific target. When including them, choose the values that most precisely identify what the feedback is about:

In the final argument, say what helped and why, or provide the relevant context, expected behavior, and what happened instead. Do not include secrets, source code, personal data, long prompts, or stack traces.

Eval candidates: tasks that broke the model

Expo turns hard real-world tasks into agent evals: anything Expo an agent can attempt — framework, EAS, tooling — qualifies, whether or not a skill was involved. The signal worth sending is a task an AI agent could not complete cleanly despite real effort: several failed attempts, a build or screen that never worked, or the user stepping in to fix it manually. Never submit quick slips the agent corrected itself, more than one candidate per session, or a task already reported.

When such a failure happens — or the user says a model failed at an Expo task — show the user the exact submission you intend to send and get approval; the Task field must describe the Expo-technical shape of the task, never the user’s product or business context. Without a user to approve it (headless or CI runs), do not submit. Then run from the failing app’s directory (the CLI attaches the SDK and package versions, agent harness, and platform automatically) with --category evals, --subject naming the Expo package or command involved (a capability phrase only when no single package fits), and this structure in the final argument:

Task: <what was asked, self-contained>.
Expected: <observable success criteria>.
Actual: <what the agent did instead>.
Wrong approach: <the specific mistake, such as a wrong API, hallucinated prop, or bad pattern>.
Evidence: <model name, attempts, how it was solved — or never was; omit what you cannot verify>.

A good candidate is solvable (eventually done or clearly doable), verifiable (success is observable), and specific. Mention only environment details the CLI cannot see, such as other key packages or a freshly created app. Describe code; do not paste it. If the command fails with an error naming evals as an invalid category, resend once with --category unknown and the same subject prefixed eval-candidate: ; on any other error, do not resend.

Usage telemetry

Automatic skill usage telemetry is separate from feedback and off by default. Never enable it without an explicit user request. Use the following script with --on, --off, or --status:

node "${CLAUDE_PLUGIN_ROOT}/skills/expo-skill-feedback/scripts/telemetry.cjs" --status

Environment equivalents are EXPO_SKILLS_TELEMETRY=1 to enable and EXPO_SKILLS_TELEMETRY=0 or DO_NOT_TRACK=1 to disable. CI never sends telemetry.

Submitting Feedback

If you encounter errors, misleading or outdated information in this skill, report it so Expo can improve:

npx --yes submit-expo-feedback@latest --category skills --subject "expo-skill-feedback" "<actionable feedback>"

Only submit when you have something specific and actionable to report. Include as much relevant context as possible. If an AI agent repeatedly failed or the user had to take over an Expo task, load the expo-skill-feedback skill and follow its eval-candidate flow instead of reusing the command above.