ReasoningBank with AgentDB
Pass
Audited by Gen Agent Trust Hub on Apr 2, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is composed of a single documentation file (SKILL.md) providing instructions for integrating with AgentDB. No executable code or malicious scripts are included in the skill distribution.
- [EXTERNAL_DOWNLOADS]: The instructions recommend using
npx agentdb@latestto initialize the database and set up the MCP server. This involves fetching the package from the standard npm registry. - [COMMAND_EXECUTION]: The documentation provides standard shell commands for database initialization, migration, and management. These commands are typical for the described functionality of setting up a local vector database.
- [PROMPT_INJECTION]: The skill describes a system that ingests user queries and agent trajectories into a database for retrieval. This creates a surface for indirect prompt injection where data processed at runtime could influence agent behavior, but the static instructions do not contain any malicious injection patterns.
Audit Metadata