space-gemini-article-illustration
Pass
Audited by Gen Agent Trust Hub on Jul 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local Python script (
scripts/generate_image.py) to perform its core image generation and file management tasks. - [EXTERNAL_DOWNLOADS]: The skill communicates with the Google Gemini API at
generativelanguage.googleapis.com. This is a well-known and trusted service required for the skill's described functionality. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted Markdown articles to generate instructions for the image model.
- Ingestion points: Reads article content from the user's local filesystem or active editor context as defined in
SKILL.mdandPhase 1instructions. - Boundary markers: None; the skill does not explicitly instruct the agent to ignore or delimit potentially malicious instructions within the source articles.
- Capability inventory: The agent can execute shell commands, perform network requests to external APIs, and write files to the local directory (
/Users/ugreen/Documents/obsidian/). - Sanitization: No evidence of input validation, sanitization, or filtering was found to prevent embedded instructions in the articles from influencing the agent's behavior.
Audit Metadata