chatgpt-image-generator
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from ChatGPT, which constitutes an external, untrusted source. 1. Ingestion points: Prompt responses, image metadata, and search results returned within the ChatGPT browser session. 2. Boundary markers: The skill instructions prescribe a 'Research-Aided Mode' and 'Campaign Co-Creation Mode' to separate data ingestion from generation actions. 3. Capability inventory: Use of the in-app browser for navigation and interaction, and file system writes for saving image files and JSON manifests. 4. Sanitization: No explicit sanitization of ChatGPT output is defined; the skill relies on the agent's internal safety guardrails and user review.
- [DYNAMIC_EXECUTION]: The workflow involves dynamic loading of helper scripts using
import()calls to../../scripts/chatgpt-bridge.mjs. This is a standard functional pattern for this skill to bridge the agent context with the browser environment using local relative paths. - [DATA_EXPOSURE_AND_EXFILTRATION]: The skill interacts with sensitive session states (login, CAPTCHA) but includes explicit instructions not to inspect, export, or log cookies, session tokens, or passwords, directing the user to handle authentication directly in the browser surface.
Audit Metadata