pexels-media
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses
curlandjqin documented workflow examples to interact with the Pexels API, process JSON responses, and download media files. These operations are standard and limited to the skill's primary purpose. - [EXTERNAL_DOWNLOADS]: Fetches media content from well-known and official Pexels domains (
images.pexels.com,videos.pexels.com). These are recognized services and the downloads are essential for the skill's stated functionality. - [DATA_EXPOSURE_EXFILTRATION]: Correctly utilizes the
PEXELS_API_KEYenvironment variable for authentication rather than hardcoding credentials. Network operations are directed exclusively to official Pexels API endpoints. - [INDIRECT_PROMPT_INJECTION]: The skill processes external data from Pexels API responses (e.g., photographer names, image descriptions, and alt text).
- Ingestion points: API response data processed in
SKILL.mdworkflows. - Boundary markers: Absent; the agent is instructed to process the raw JSON output.
- Capability inventory: File writing (creating
.meta.jsonfiles) and file downloads viacurl. - Sanitization: None explicitly defined for prompt interpolation; however, the risk is minimal given the trusted nature of the content provider.
Audit Metadata