ads-generate
Pass
Audited by Gen Agent Trust Hub on Mar 13, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes Python code directly using
python3 -cin Step 1 to verify that required environment variables (e.g., GOOGLE_API_KEY) are set. - [COMMAND_EXECUTION]: In standalone mode, the skill invokes a local script
~/.claude/skills/ads/scripts/generate_image.pyvia the command line to generate images, passing user-provided prompts as arguments. - [PROMPT_INJECTION]: The skill processes untrusted data from
campaign-brief.mdandbrand-profile.json, which are used to instruct thevisual-designeragent and provide parameters for image generation. - Ingestion points: Content is read from
campaign-brief.mdandbrand-profile.json(referenced in SKILL.md). - Boundary markers: Absent; there are no explicit delimiters or instructions to the sub-agents to ignore potentially malicious instructions embedded within the brief files.
- Capability inventory: The skill possesses the capability to execute subprocesses (Python scripts), write files to the local directory structure (
./ad-assets/), and spawn additional AI agents. - Sanitization: Absent; the instructions do not specify any validation or sanitization steps for the input file content before it is used in prompt interpolation or shell commands.
Audit Metadata