minimax-delegation
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests local file content to provide context for AI tasks, which can be exploited via indirect prompt injection if the files contain adversarial instructions.\n
- Ingestion points: delegation_executor.py reads files from the local filesystem to include in the model prompt.\n
- Boundary markers: Not specified; file content is inlined directly into the command arguments.\n
- Capability inventory: The skill can read local files, execute shell commands via mmx, and transmit data to the MiniMax API.\n
- Sanitization: No sanitization or escaping of the inlined file content is mentioned in the documentation.\n- [EXTERNAL_DOWNLOADS]: The skill directs the user to install the mmx-cli package globally via npm from its official repository. This is the legitimate tool for interacting with MiniMax services.\n- [COMMAND_EXECUTION]: The skill relies on executing the mmx binary and a local Python helper script (delegation_executor.py) to perform its tasks.
Audit Metadata