AgentDB Advanced Features

Pass

Audited by Gen Agent Trust Hub on Aug 21, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill features a context synthesis capability that generates narratives from retrieved database entries. Ingestion points: Data is fetched via the adapter.retrieveWithReasoning method described in SKILL.md. Boundary markers: There are no delimiters or instructions to the agent to ignore potentially malicious embedded content in the retrieved data. Capability inventory: The agent has access to network synchronization and CLI-based database operations. Sanitization: No sanitization of the retrieved data is demonstrated before it is provided as context for synthesis.
  • [UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The documentation specifies the use of npx agentdb@latest for various database tasks, which downloads and executes the latest version of the package from the public registry at runtime.
  • [DATA_EXPOSURE_AND_EXFILTRATION]: The QUIC synchronization feature allows the database to transmit patterns across the network to defined peers. If directed to untrusted addresses via syncPeers or the AGENTDB_QUIC_PEERS environment variable, this could result in the unauthorized exfiltration of stored database content.
  • [PRIVILEGE_ESCALATION]: Troubleshooting sections provide commands using sudo ufw to modify system firewall settings, which is a privileged operation that requires human verification before execution.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 21, 2026, 05:04 AM
Security Audit — agent-trust-hub — AgentDB Advanced Features