Looking Up Members

Resolution Algorithm

Given a query string (name or email fragment):

1. Exact email match: compare query (case-insensitive) against org_members[*].email. Return immediately if exactly one match.
2. Exact name match: compare query (case-insensitive) against org_members[*].name. Return immediately if exactly one match.
3. Partial name match: check if any name contains the query string (case-insensitive). Collect all matches.
4. Return results:
   • 0 matches → return empty array []. Do not error — let the caller handle the missing case.
   • 1 match → return [{ id, name, email }].
   • 2–5 matches → return the array. Caller should present candidates and ask the user to confirm.
   • 6+ matches → return the first 5 with a note that the query is too broad. Caller should ask for a more specific name.

TeamsDB Fallback

If org_members is empty (provider does not support member listing), fall back to TeamsDB:

1. If teamsDatabaseId is not in headless_config, return empty array [] (no fallback available).
2. Fetch the Teams database using teamsDatabaseId from config.
3. Collect all unique persons from the Members (people-type) field across all teams.
4. Use this as the search corpus. Cache result in org_members for the session.
5. Apply the same Resolution Algorithm above.

Caller Responsibility

The caller (delegating-tasks, ingesting-messages, managing-tasks) must:

• Run detecting-provider + resolving-identity (org_members) before calling this skill.
• Handle the returned candidates with AskUserQuestion if multiple matches exist.
• Not call this skill for the current user — use current_user.id directly from resolving-identity.