fathom-core-workflow-a
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external meeting transcripts and summaries, creating a potential attack surface where malicious instructions hidden in meeting content could attempt to influence the agent's behavior.
- Ingestion points: Data is ingested through the
client.list_meetings()method in the Python snippets withinSKILL.md. - Boundary markers: The skill does not implement specific delimiters or instructions to treat meeting content as untrusted data.
- Capability inventory: The skill is allowed to use
Bash(python3:*),Bash(curl:*),Read,Write,Edit, andGrep. - Sanitization: There is no logic provided to sanitize or filter the meeting content before it is processed or printed.
Audit Metadata