spark-recipe-meeting-followup
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes meeting transcripts and notes that may contain untrusted data or hidden instructions from meeting participants. If the agent processes a transcript containing malicious instructions, it could be influenced to deviate from the user's intent when drafting emails or posting comments.
- Ingestion points: Meeting data fetched via
spark meeting <id> --transcript --notesin SKILL.md. - Boundary markers: The instructions do not specify the use of delimiters or 'ignore instructions' warnings when handling the transcript data.
- Capability inventory: The skill uses
spark draftto create emails andspark commentto post on threads based on the ingested content. - Sanitization: No sanitization, validation, or escaping of the transcript content is instructed before it is interpolated into email bodies.
Audit Metadata