google-business-profile
Pass
Audited by Gen Agent Trust Hub on Nov 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local Python script named
brand_context.pylocated at../start-here/scripts/. Invoking executable code from a relative path outside the skill's immediate directory is a functional risk that relies on the integrity of the sibling folder structure. - [INDIRECT_PROMPT_INJECTION]: The skill processes data from untrusted external sources, specifically competitor websites and public Google Business Profiles. This content could be crafted by an adversary to influence the agent's audit logic or social media post generation.
- Ingestion points: Competitor site content and public profile fields are retrieved using web search tools during the research phase.
- Boundary markers: The instructions do not define the use of markers or delimiters to isolate scraped external text from the agent's core instructions.
- Capability inventory: The skill possesses the ability to write local audit files, execute local scripts, and interact with external publishing and scraping APIs.
- Sanitization: The skill mandates human screening for photo content but lacks explicit sanitization or validation protocols for ingested text content.
- [DATA_EXFILTRATION]: The skill facilitates network operations to external services such as Zernio and Firecrawl. While these are documented tool integrations used for scraping and publishing, they involve transmitting business-related data and metadata to non-whitelisted third-party domains.
Audit Metadata