top-k-vector-pruning
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill implements a data processing pipeline for content retrieved from external vector databases. While this involves ingesting potentially untrusted data, the skill serves as a security-enhancing filter by restricting the volume of external data based on similarity thresholds (Score >= 0.78) and Top-K clamping (max 2 or 3 chunks), thereby reducing the attack surface for indirect prompt injection.
- [REMOTE_CODE_EXECUTION]: The included Python script performs standard list operations and mathematical comparisons. It does not invoke subshells, execute dynamic strings, or fetch remote code.
- [DATA_EXFILTRATION]: No network operations or commands capable of accessing sensitive system files (e.g., credentials, SSH keys) were detected in the instructions or the code.
Audit Metadata