dev-verify
Pass
Audited by Gen Agent Trust Hub on Jun 22, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes hooks to execute local Python scripts (
dev-delegation-guard.pyandphase-gate-guard.py) viauv runto enforce security policies and phase gates before sensitive tools are used. - [COMMAND_EXECUTION]: Employs dynamic context injection (
!uv run ...) to load operational constraints from a local path (scripts/load-constraints.py) during the skill loading phase. - [PROMPT_INJECTION]: The skill processes project requirements from files such as
.planning/SPEC.mdand.planning/PLAN.mdto verify feature completion. While this constitutes an indirect prompt injection surface, the risk is mitigated by the use of a read-only subagent for this task and the absence of network exfiltration capabilities. - [COMMAND_EXECUTION]: Spawns a subagent (
workflows:dev-verifier) with restricted, read-only tools to perform goal-backward verification without the ability to modify the codebase.
Audit Metadata