bmad-forge-idea
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes
uv runto execute a series of local Python scripts for system orchestration and state management. These include core scripts (resolve_customization.py,resolve_config.py,memlog.py) and a local persona resolver (resolve_personas.py). These operations are used to handle configuration, logging, and environment setup within the project structure.\n- [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it processes external project materials and configurable facts to inform its questioning strategy. \n - Ingestion points: File contents loaded via
persistent_factsincustomize.tomland project-specific documents provided by the user during the session.\n - Boundary markers: Absent; the skill integrates these contents directly into the conversation context.\n
- Capability inventory: The agent can write state logs and reports to the file system and execute local Python orchestration scripts.\n
- Sanitization: No specific sanitization or filtering logic is applied to the ingested external content before it is used by the LLM.\n- [SAFE]: Analysis of the skill's code and instructions found no evidence of malicious intent, data exfiltration to external domains, or unauthorized persistence mechanisms. All script executions and file operations are consistent with the stated purpose of idea development and project management.
Audit Metadata