qcsd-development-swarm
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill analyzes external source code and test files provided via SOURCE_PATH and TEST_PATH, creating a potential surface for indirect prompt injection. \n
- Ingestion points: User-provided source code and test directories. \n
- Boundary markers: None identified in the provided step files. \n
- Capability inventory: Use of the Read tool for ingestion, Workflow and Task tools for agent spawning, and file system writes for reports. \n
- Sanitization: No specific sanitization or filtering of code content is documented before analysis. \n- [COMMAND_EXECUTION]: The skill utilizes a domain-specific CLI tool 'aqe' to search and store state data (e.g., 'aqe memory search' in Step 1 and 'aqe memory store' in Step 7). This represents legitimate use of custom tooling for swarm orchestration.
Audit Metadata