token-saver-adaptive-ratio

Pass

Audited by Gen Agent Trust Hub on Jul 5, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill performs mathematical calculations to determine compression ratios (0.8, 0.5, 0.2, or 0.1) based on an input token count. No malicious patterns detected.
  • [SAFE]: The script scripts/main.cjs utilizes require() to load logic from a relative directory path (../../token-saver-context-compression/scripts/main.cjs). While this creates a dependency on the external file structure, it is a standard module resolution pattern for related tools from the same author.
  • [PROMPT_INJECTION]: The skill defines an input schema that accepts a context object for analysis. 1. Ingestion points: context in schemas/input.schema.json. 2. Boundary markers: Absent. 3. Capability inventory: No critical capabilities (network, file-write, or subprocess) found across scripts. 4. Sanitization: Absent. As the skill only calculates a numeric ratio based on token count, the impact of indirect injection is negligible.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 5, 2026, 10:58 AM
Security Audit — agent-trust-hub — token-saver-adaptive-ratio