morning-brief
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes potentially untrusted data from external sources to generate automated emails and summary reports.
- Ingestion points: Processes lead activity and contact details from HubSpot CRM, meeting notes and action items from Clari call summaries (via
clari_get_call_summary), and event descriptions from Google Calendar. - Boundary markers: The workflow lacks explicit delimiters or "ignore previous instructions" warnings when interpolating external call summaries or CRM data into the Gmail draft templates or the HTML briefing document.
- Capability inventory: The skill possesses significant capabilities, including reading sensitive CRM/Calendar/Call data, creating Gmail drafts (
gmail_create_draft), and writing telemetry to the local filesystem (~/.claude/skill-analytics/). - Sanitization: There is no evidence of sanitization, escaping, or filtering of external content (e.g., call summaries which might be influenced by meeting participants) before it is used to drive the agent's generation of email drafts.
Audit Metadata