AgentDB Memory Patterns

Pass

Audited by Gen Agent Trust Hub on Aug 21, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill uses npx agentdb@latest to fetch the AgentDB command-line tool from the official NPM registry.
  • [REMOTE_CODE_EXECUTION]: By utilizing npx, the skill executes remote code from the NPM registry to initialize databases and manage plugin templates.
  • [COMMAND_EXECUTION]: The documentation provides commands for adding an MCP (Model Context Protocol) server to the agent environment, enabling persistent database interactions.
  • [PROMPT_INJECTION]: The skill implements persistent memory management that processes and stores interaction data, creating a surface for potential indirect prompt injection attacks. 1. Ingestion points: Memory storage operations such as storeMemory, storeFact, and insertPattern as documented in SKILL.md. 2. Boundary markers: The integration patterns do not include explicit delimiters or instructions for the agent to treat retrieved data as untrusted content. 3. Capability inventory: The skill makes use of shell execution for CLI operations and requires file system access for database persistence. 4. Sanitization: No evidence of data validation or input filtering is present in the provided implementation examples.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 21, 2026, 05:04 AM
Security Audit — agent-trust-hub — AgentDB Memory Patterns