Post-install hint

When gbrain integrations install retrieval-reflex --target <repo> completes, print this to stdout (or to the install agent's conversation surface) so the operator knows what to do next.

---

✓ Retrieval-reflex policy skill installed to <target-repo>/skills/retrieval-reflex/.

Three follow-up steps:

1. Confirm the deterministic pointer layer is healthy

gbrain doctor --json | jq '.checks[] | select(.name=="retrieval_reflex_health")'

The pointer layer lives in the gbrain context engine (zero-LLM, on by default) — nothing to install. If the check reports it disabled, the policy skill alone still carries the behavior; the check output says which.

2. Review the appended resolver row

This install appended one row to your <target>/RESOLVER.md (or AGENTS.md):

retrieval-reflex | a named person/company/project/place becomes the subject mid-conversation; ...

The row is deliberately REFLEX-ONLY. Explicit lookup questions ("who is X", "tell me about Y", "what do we know about Z") belong to your query/search skill — do not add those phrases to this row, or the two skills will contend for the same requests and routing becomes ambiguous.

3. Update later

When gbrain ships a new retrieval-reflex policy, refresh your local copy:

gbrain integrations install retrieval-reflex --target <target-repo> --refresh

Local edits are preserved by default; pass --auto take-theirs to take upstream everywhere, or --dry-run to preview.