AgentDB Performance Optimization

Pass

Audited by Gen Agent Trust Hub on Aug 21, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill instructs the use of npx agentdb@latest, which dynamically downloads and executes code from the npm registry at runtime without version pinning.
  • [COMMAND_EXECUTION]: Documentation provides shell commands (npx agentdb benchmark and npx agentdb stats) for the agent to execute to measure database performance.
  • [EXTERNAL_DOWNLOADS]: The skill references and links to external codebases and documentation hosted on GitHub (github.com/ruvnet/agentic-flow) and an external domain (agentdb.ruv.io).
  • [PROMPT_INJECTION]: The skill processes external data via vector database lookups using retrieveWithReasoning, creating a surface for indirect prompt injection.
  • Ingestion points: adapter.retrieveWithReasoning and adapter.insertPattern methods in SKILL.md.
  • Boundary markers: None provided in the examples to distinguish between data and instructions.
  • Capability inventory: The agent has capabilities to execute shell commands via npx and modify local database files.
  • Sanitization: No explicit sanitization or validation of the retrieved vector content is mentioned before it is processed by the reasoning engine.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 21, 2026, 05:03 AM
Security Audit — agent-trust-hub — AgentDB Performance Optimization