subagent-driven-development
Pass
Audited by Gen Agent Trust Hub on Aug 31, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill performs various shell operations for workflow management, including
git status,git diff, andgit rev-parse. It also invokes a local Python scriptscripts/check-scope-overlap.pyto manage task dependencies. - [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to instructions embedded in project data. Ingestion points: The skill extracts task text from external plan files (SKILL.md, Phase 1) and Architectural Decision Records (ADRs). Boundary markers: The workflow does not utilize explicit delimiters or instructions to ignore embedded commands within the extracted task descriptions. Capability inventory: The agent and its dispatched subagents have access to powerful tools including
Bash,Edit,Write, andTask(subagent creation). Sanitization: There is no evidence of validation or sanitization performed on the ingested plan content before it is interpolated into prompts.
Audit Metadata