caveman-marketing
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data (URLs, screenshots, or pasted copy) which may contain hidden instructions designed to influence the agent's behavior.
- Ingestion points: The
SKILL.mdinstructions define inputs derived from external URLs, screenshots, and raw text copy from ads or landing pages. - Boundary markers: The skill does not provide specific instructions for using delimiters or boundary markers to isolate untrusted content from the agent's core instructions.
- Capability inventory: The skill utilizes the agent's text generation capabilities and instructs the agent to execute a local Python validation script (
scripts/check.py). - Sanitization: The skill does not implement or recommend sanitization, filtering, or escaping of the input content before processing.
- [COMMAND_EXECUTION]: The skill includes and instructs the agent to execute a local Python script (
scripts/check.py) and a shell test script (scripts/test.sh) for internal validation. - The included scripts use standard Python libraries for text analysis and standard shell utilities for testing.
- The execution is scoped to the local environment for validating agent-generated output against formatting rules.
Audit Metadata