- Add user_access_token support to api_get/api_post for user-identity API calls
- Add fetch_p2p_messages() to collect both sides of a private conversation
- Extend collect_messages() to combine p2p + group chat messages
- Add --exchange-code to convert OAuth code to user_access_token
- Add --user-token, --p2p-chat-id, --open-id CLI flags
- Update SKILL.md with p2p collection flow and permission requirements
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The old find_user() used /search/v1/user which requires user_access_token,
but the code only has tenant_access_token, causing error 99991663 every time.
New strategy:
1. Email/phone → /contact/v3/users/batch_get_id (fastest, proven working)
2. Name → department traversal + find_by_department (needs contact:department.base:readonly)
3. Both fail → clear error message with suggestions
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Merge 7 intake questions into 3: name, basic info, personality profile
- Each question includes a one-shot example
- Change "姓名" to "花名/代号", slug uses - separator
- Unify all references from colleague-creator to create-colleague
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
