openpress-plugins
Pass
Audited by Gen Agent Trust Hub on Aug 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill identifies and recommends the installation of external agent skills from GitHub repositories belonging to individual developers (
cathrynlavery/diagram-designandLiamGvchi/gc-minimal-zine-poster). These sources are not recognized as well-known technology organizations or trusted services. - [PROMPT_INJECTION]: The skill processes content and outputs from external plugins to generate functional React components (e.g.,
ArchitectureDiagramFigure.tsxandCoverPoster.tsx). This mechanism is vulnerable to indirect prompt injection if an external plugin provides malicious data that is then interpolated into the generated code. - Ingestion points: User intent matching (triggers) and data returned by external specialized skills (e.g., diagram data, visual descriptions).
- Boundary markers: None identified. There are no explicit instructions for the agent to use delimiters or ignore instructions embedded within the external plugin's output.
- Capability inventory: File system writes (creating components and media files), invocation of other agent skills, and interpolation of data into executable React/TypeScript templates.
- Sanitization: None identified. The adapters directly use external data to fill in React component props and template literals without explicit escaping or validation.
Audit Metadata