bmad-party-mode
Pass
Audited by Gen Agent Trust Hub on May 11, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill processes project-specific context files to dynamically build prompts for subagents.
- Ingestion points: It reads
agent-manifest.csvandproject-context.mdfrom the project directory to define subagent identities and provide discussion background. - Boundary markers: The skill uses structured Markdown headers (e.g., '## Your Persona', '## Discussion Context') to delineate persona data and project context from internal instructions.
- Capability inventory: The skill uses the 'Agent' tool to execute subagents. It specifically mitigates risks by instructing all spawned subagents not to use any tools, limiting the potential impact of any instructions embedded in the project files.
- Sanitization: It implements context management by summarizing conversation history to fit within a 400-word limit before passing it to subagents, which acts as a basic form of content filtering.
Audit Metadata