party-mode
Pass
Audited by Gen Agent Trust Hub on Aug 8, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill executes internal Python scripts (
resolve_party.py,memlog.py) to handle configuration merging and memory logging. These scripts use Python's standard library and perform local file operations within the skill's scope. - [EXTERNAL_DOWNLOADS]: The skill utilizes the agent's native web-search tools to conduct research for Tier 2 and Tier 3 discussions. It does not download or execute remote scripts or binaries.
- [DATA_EXFILTRATION]: Data management is confined to the workspace. Memory files and reports are stored in designated local directories (
docs/party-mode-memories/anddocs/party-mode-report/). No unauthorized network transmission of sensitive data was identified. - [PROMPT_INJECTION]: The skill defines various personas, including a security engineer and an adversary, to provide diverse perspectives during discussions. These are character roles intended for the discussion context and do not attempt to override the underlying AI's safety protocols.
- [REMOTE_CODE_EXECUTION]: No patterns of remote code execution, such as piping web content to a shell, were found. The skill relies on its bundled scripts and standard environment tools.
Audit Metadata