agent-brief
Pass
Audited by Gen Agent Trust Hub on Nov 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill implements a robust "data vs instructions" policy in
pipeline.md, specifically instructing the agent to ignore any executable commands found in project files, designs, or ADRs. - [SAFE]: The bundled validation script
scripts/brief.pyuses only Python standard libraries and is invoked using the-I(isolated) flag, which prevents environment contamination and protects against script-based injection. - [INDIRECT_PROMPT_INJECTION]: The skill has an ingestion surface for untrusted data as it reads designs, ledger files, and codebase "prior art" from the local repository. This risk is addressed through strict instructions in
pipeline.mdanddelegation.mdto treat ingested content as data rather than instructions, and by requiring an explicit user approval gate before any build plan is finalized.
Audit Metadata