ad-management-intake
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data which could contain malicious instructions designed to influence the agent's behavior.
- Ingestion points: External brand website content accessed via the research step (references/gooseworks-adapter.md), user-provided notes or documents (references/local-adapter.md), and direct user chat messages.
- Boundary markers: Absent. The skill does not specify the use of delimiters or instructions to ignore embedded commands within the processed data.
- Capability inventory: The skill performs file writes to the 'ads/' directory, updates campaign records via the 'campaign_upsert' tool, and executes a shell command for validation.
- Sanitization: Absent. There is no mention of sanitizing or validating external input before it is used to populate internal fact sheets or campaign briefs.
- [COMMAND_EXECUTION]: The skill instructions include a command to execute a local Python script for output validation.
- Evidence: The 'references/local-adapter.md' file instructs the agent to run 'python3 /contract/validate.py ads'. While this targets a script within a required peer skill ('meta-ad-manager'), it represents the execution of a shell command based on a relative path.
Audit Metadata