write-script-bunnative
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to use various CLI commands (wmill script preview, wmill script run, wmill generate-metadata) to interact with the Windmill platform. This is core functionality for developer workflows as defined by the vendor, windmill-labs.
- [EXTERNAL_DOWNLOADS]: The instructions permit importing npm packages within the Bun Native scripts. These dependencies are automatically resolved by the platform's build system and are consistent with the intended use of the Bun runtime.
- [INDIRECT_PROMPT_INJECTION]: The skill documents preprocessor scripts that ingest untrusted data (e.g., webhooks, SQS, email) into the agent context in SKILL.md. No specific boundary markers or 'ignore' instructions are provided in the guide. The capability inventory for such scripts includes subprocess execution via the wmill CLI, network operations via fetch, and S3 file access. Sanitization or escaping guidelines for external content are not explicitly detailed in this documentation.
Audit Metadata