write-swarm-client
Pass
Audited by Gen Agent Trust Hub on May 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to create directories and multiple Python source files (e.g., agent_roll.py, agent_run.py) in the local filesystem to establish a training environment.
- [PROMPT_INJECTION]: The skill facilitates the creation of agents that process untrusted task data, which creates a potential surface for indirect prompt injection attacks. 1. Ingestion points: External data enters the agent through the
task.main_queryfield. 2. Boundary markers: The code templates do not implement delimiters or instructions to ignore instructions embedded within the task data. 3. Capability inventory: The generated agents are capable of calling large language models via the OpenAI SDK and executing Python code for task completion. 4. Sanitization: The skill does not provide any logic for sanitizing or validating input data before processing.
Audit Metadata