delegate-brief
Pass
Audited by Gen Agent Trust Hub on Nov 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill framework involves generating and executing shell commands (such as Node.js scripts, build commands, and testing tools) within sub-agent briefs to facilitate automated development tasks. This is a functional requirement of the delegation workflow. Evidence in
SKILL.mdandsentinel-brief-template.md. - [INDIRECT_PROMPT_INJECTION]: The delegation process ingests project files (specs, plans, and code) to provide context to sub-agents. The skill mitigates injection risks by requiring mandatory structural validation via
scripts/check-brief.mjs, defining explicit action boundaries, and maintaining a list of forbidden adversarial instructions (e.g., "challenge design" or "unauthorized probing") for review roles. - [SAFE]: The included JavaScript files (
scripts/check-brief.mjsandscripts/lib/check-brief.mjs) are legitimate utility scripts designed for markdown template validation and do not perform any hidden or dangerous operations.
Audit Metadata