AgentDB Learning Plugins
Pass
Audited by Gen Agent Trust Hub on Mar 4, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the user to use
npx agentdb@latest, which downloads and executes the vendor's package from the NPM registry. - [COMMAND_EXECUTION]: Provides multiple shell commands for the
agentdbCLI to create, list, and manage learning plugins. - [PROMPT_INJECTION]: The skill exhibits vulnerability to indirect prompt injection (Category 8):
- Ingestion points: The
adapter.insertPatternmethod inSKILL.mdallows the storage of external 'experience' data. - Boundary markers: None identified; the skill does not suggest delimiters or instructions to ignore embedded commands in training data.
- Capability inventory: The ingested data is utilized by
adapter.trainandadapter.retrieveWithReasoningto influence agent decision-making and context synthesis. - Sanitization: No evidence of input validation or sanitization of the
pattern_datafield is provided in the examples.
Audit Metadata