qdrant-performance-optimization

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill contains multiple references to the official Qdrant technical documentation and articles hosted at skills.qdrant.tech. These links are vendor-provided resources used to supplement the troubleshooting guidance for users.
  • [INDIRECT_PROMPT_INJECTION]: The skill implements a routing table that uses user-provided symptoms to determine which local markdown files the agent should read. This creates a surface where user input influences the agent's context; however, the agent's capabilities are limited by the platform configuration to basic file reading tools (Read, Grep, Glob), and the content being ingested consists of static documentation provided within the skill package.
  • Ingestion points: User symptoms mapped to specific file paths in SKILL.md.
  • Boundary markers: None explicitly defined in the routing logic.
  • Capability inventory: The skill is limited to Read, Grep, and Glob tools as defined in the frontmatter.
  • Sanitization: Not applicable as the skill reads static documentation files based on pre-defined mappings.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 10:33 PM
Security Audit — agent-trust-hub — qdrant-performance-optimization