AgentDB Learning Plugins
Pass
Audited by Gen Agent Trust Hub on Feb 28, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill uses
npx agentdb@latestand references theagentic-flowlibrary, which requires downloading packages from the public NPM registry at runtime. - [COMMAND_EXECUTION]: Multiple CLI examples provided in the documentation use
npx agentdb@latest, which facilitates the execution of code from a remote NPM package to manage plugins. - [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface through its training data ingestion and retrieval processes. Malicious instructions embedded in the 'experience' data could influence agent decisions during reasoning phases.
- Ingestion points: Training data is ingested via the
adapter.insertPatternmethod inSKILL.md. - Boundary markers: No specific delimiters or instructions to ignore embedded commands were identified in the provided code snippets.
- Capability inventory: The skill is capable of training models and performing reasoning-based retrieval (
adapter.train,adapter.retrieveWithReasoning). - Sanitization: There is no evidence of sanitization or validation of the
pattern_datafield before it is stored or used for training.
Audit Metadata