llm-cost-optimization
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references and downloads pre-trained models for semantic similarity and prompt compression, specifically
BAAI/bge-small-en-v1.5andmicrosoft/llmlingua-2-bert-base-multilingual-cased-meetingbank. These are provided by well-known research and technology organizations. - [INDIRECT_PROMPT_INJECTION]: The skill provides patterns for processing external data (e.g., long context files and user prompts) which creates an entry point for untrusted content.
- Ingestion points:
open("large-context.txt").read()and thetextslist processing inSKILL.md. - Boundary markers: No explicit delimiters or instructions to ignore embedded commands are included in the provided snippets.
- Capability inventory: The skill uses network-connected tools (Redis, LLM APIs) and file system read/write operations.
- Sanitization: No data sanitization or input validation is present in the logic. This represents a standard vulnerability surface for skills processing external text, but is assessed as safe in the context of these cost optimization templates.
Audit Metadata