unsloth
Pass
Audited by Gen Agent Trust Hub on Nov 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill consists exclusively of documentation, references, and usage guidelines for the Unsloth fine-tuning framework. It contains no executable scripts, shell command instructions, or active network components.\n- [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to provide technical documentation derived from external sources to the agent. This creates a surface for indirect prompt injection where instructions in the source text could influence the agent, though no such malicious patterns were identified in the current reference files.\n
- Ingestion points: Data enters the context via
references/llms.mdandreferences/index.md.\n - Boundary markers: The content is organized with markdown headers but lacks explicit safety delimiters to isolate embedded instructions.\n
- Capability inventory: No tool-use capabilities, subprocess calls, or file/network operations are defined in the skill.\n
- Sanitization: None detected.
Audit Metadata