The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to fetch configuration settings and PII/PHI detection audit logs from the platform API.
Evidence: GET requests to api.iblai.app/dm/api/ai-mentor/orgs/{org}/users/{username}/mentors/{mentor}/settings/ and .../privacy-flags/.
[COMMAND_EXECUTION]: Includes template instructions for the agent to execute curl commands to interact with the API.
Evidence: curl -X PUT examples provided for updating privacy actions, entities, and response strings.
[DATA_EXFILTRATION]: Transmits privacy configuration updates and state changes to the vendor's backend infrastructure.
Evidence: PUT requests sending fields such as enable_privacy_router, privacy_action, and privacy_entities to api.iblai.app.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from an audit log which may contain untrusted user-supplied fields.
Ingestion points: Data ingested via GET requests to the privacy-flags endpoint described in SKILL.md.
Boundary markers: No explicit markers or validation rules are defined for the agent when interpreting the API response body.
Capability inventory: The skill provides capabilities for authenticated API reads (audit logs) and writes (settings updates) using curl.
Sanitization: No specific sanitization or filtering logic is documented for fields like username, session_id, or search parameters returned by the API.

iblai-api-agent-privacy