bro-give-me-spec
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [SAFE]: The skill implements a secure document generation workflow that focuses exclusively on desired outcomes rather than implementation details. By explicitly forbidding the generation of code, diffs, or step-by-step implementation plans, the skill mitigates the risk of introducing malicious logic into a codebase.
- [SAFE]: Strong access controls are enforced for file operations. The skill is prohibited from selecting its own storage paths and must obtain explicit user approval before overwriting any existing files, protecting against unauthorized data modification.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses an ingestion surface for untrusted data as it reads external code and user requirements (SKILL.md, Step 1). This is mitigated by mandatory boundary markers in the form of structural templates (spec-template.md) and a dedicated verification loop using a read-only sub-agent (closer-prompt.md). The capability inventory includes writing to local files and calling platform-specific tools, but the risk is lowered by the prompt's focus on observable criteria rather than executable instructions.
- [COMMAND_EXECUTION]: The skill invokes platform-integrated tools such as
CreatePlan,AskQuestion, and/bro-review-spec-plan. These appear to be authorized environment features designed for managed planning and review workflows rather than arbitrary shell execution.
Audit Metadata