recoup-internal-consulting-email-atomizer
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted external data, such as industry events and call transcripts, to generate email drafts. This ingestion of arbitrary text creates an attack surface where malicious instructions embedded in the source material could influence the agent's drafting or routing logic.
- Ingestion points: Source content is pulled from signals files, raw call transcripts, and external industry event descriptions as described in SKILL.md Step 1.
- Boundary markers: The instructions lack specific delimiters or instructions for the agent to treat the source content as untrusted data or to ignore embedded instructions.
- Capability inventory: The skill is capable of writing files to the local workspace (email/outbox/) and executing a local Gmail integration script (integrations/gmail/_work/create_draft.py).
- Sanitization: The skill employs a verification gate (the recoup-internal-consulting-outbound-email skill) to review all drafts before they are staged.
- [COMMAND_EXECUTION]: The skill executes a local Python script integrations/gmail/_work/create_draft.py to stage Gmail drafts. This execution is safeguarded by an explicit requirement for user confirmation before the script is run.
Audit Metadata