where-to-start-situation-save
Pass
Audited by Gen Agent Trust Hub on Nov 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data provided by the user (founder answers) and content read from external websites to update brand records and workspace files.
- Ingestion points: Ingests 'founder's answers' and site observations into the conversation context to format the
brand/situation.mdfile. - Boundary markers: The instructions define a specific markdown structure for the output file but do not specify clear delimiters or warnings for the agent to ignore instructions potentially embedded in the user's answers or site data.
- Capability inventory: Utilizes
gooseworks:file_readfor reading existing situation data andgooseworks:brand_updatefor modifying brand profile fields. It also invokes thesave-to-brainskill to persist findings. - Sanitization: The skill provides formatting rules (e.g., separating founder words from agent guesses) but lacks explicit sanitization or validation of the content before it is written to the brand profile or workspace files.
Audit Metadata