payload
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADS
Full Analysis
- [EXTERNAL_DOWNLOADS] (LOW): The skill explicitly instructs the AI agent to fetch remote TypeScript type definitions from GitHub (e.g.,
https://github.com/payloadcms/payload/blob/main/...). While intended for developer assistance, fetching external, non-static content from a source outside the provided trusted organizations list can allow untrusted data to influence agent behavior (Indirect Prompt Injection). - [DATA_EXFILTRATION] (SAFE): The skill follows security best practices by using environment variables (e.g.,
process.env.S3_SECRET_ACCESS_KEY) in its code examples instead of hardcoding credentials. No unauthorized network requests or exfiltration patterns were found. - [COMMAND_EXECUTION] (SAFE): No dangerous shell commands, subprocess calls, or unauthorized script executions were identified in the reference files.
- [PROMPT_INJECTION] (SAFE): The instructional text is descriptive and does not contain attempts to override system prompts or bypass safety filters.
Audit Metadata