llama-factory
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill incorporates extensive documentation extracts that the agent uses to generate shell commands and Python scripts for model training.
- Ingestion points: Reference files in the
references/directory containing LLaMA-Factory documentation. - Boundary markers: None identified in the markdown structure beyond standard headers.
- Capability inventory: Documentation covers shell command execution (
llamafactory-cli,torchrun), package management (pip), and Python code execution for ML tasks. - Sanitization: Not applicable as the content is static documentation from official project sources.
- [EXTERNAL_DOWNLOADS]: The skill references installation instructions for various ML packages and the LLaMA-Factory tool itself from official sources like PyPI and Hugging Face.
- [COMMAND_EXECUTION]: The documentation provides examples of shell commands for distributed training, model merging, and API deployment. These are presented as instructional examples for the user/agent rather than commands executed by the skill load process itself.
Audit Metadata