bmad-party-mode

Pass

Audited by Gen Agent Trust Hub on Aug 30, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill's Python script ‘scripts/resolve_party.py’ executes internal configuration scripts using ‘subprocess.run’. These calls are restricted to project-local paths and use the system's Python interpreter to resolve agent and workflow metadata.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes history from party memory logs and user-supplied data to generate and refine agent personas. This data is incorporated into agent prompts, which could allow instructions from past sessions or untrusted inputs to influence the agent's behavior.
  • Ingestion points: Party memory files (‘.memlog.md’) and external profile data (files or pasted text) processed for persona creation.
  • Boundary markers: The agent is instructed to ‘distill’ content into compact briefs and maintain persona consistency, providing a behavioral constraint rather than a hard security boundary.
  • Capability inventory: The skill can execute local scripts for configuration, read and append to memory files, and trigger configuration updates via the ‘bmad-customize’ tool.
  • Sanitization: No programmatic sanitization is applied to the memory or profile data before processing by the LLM.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 30, 2026, 10:42 PM
Security Audit — agent-trust-hub — bmad-party-mode