implementing-llms-litgpt
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill installs the litgpt library and downloads model weights from repositories belonging to trusted organizations like Microsoft, Meta, and Google. These operations target reputable domains and are consistent with standard machine learning development practices.
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of external data for model fine-tuning and pretraining. * Ingestion points: Data entering through JSON datasets (e.g., data/my_dataset.json) and text corpora as described in SKILL.md and references/training-recipes.md. * Boundary markers: None identified; instructions do not specify delimiters to separate training data from system prompts. * Capability inventory: Use of CLI tools (litgpt) and shell commands to process data and manage training jobs. * Sanitization: The data is processed through standard tokenizers without specific content filtering, which is typical for the intended use case of LLM training.
Audit Metadata