bmad-party-mode
Pass
Audited by Gen Agent Trust Hub on Nov 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted external data to generate personas and conversation context, which could be exploited to override agent behavior.
- Ingestion points: The skill ingests 'source data' such as profiles and survey exports in
references/create-party.mdand reads session history from.memlog.mdfiles inreferences/party-memory.md. - Boundary markers: The instructions do not specify explicit structural delimiters or instructions for the model to ignore embedded commands when processing raw input data.
- Capability inventory: The skill has access to
subprocess.run(viascripts/resolve_party.py), triggers file writes (viabmad-customize), and uses web search tools. - Sanitization: No explicit sanitization, filtering, or validation logic is defined for the ingested data.
- [COMMAND_EXECUTION]: The skill executes local Python scripts as part of its normal activation and state management flows.
- Evidence:
scripts/resolve_party.pyusessubprocess.run()to executeroster.py,resolve_config.py, andresolve_customization.py. - Evidence:
references/party-memory.mdinvokesmemlog.pyto append session data to memory logs. - [EXTERNAL_DOWNLOADS]: The skill recommends installing dependencies from the vendor's repository if they are missing from the environment.
- Evidence:
SKILL.mdinstructs the user to runnpx skills add bmad-code-org/BMAD-METHOD --skill bmadto install the base framework.
Audit Metadata