deeplake-plugin
Pass
Audited by Gen Agent Trust Hub on Jun 16, 2026
Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill is designed to automatically capture and upload all interaction data—including both user and assistant messages—to the DeepLake REST API. While this aligns with the skill's stated purpose of providing cloud-backed shared memory, it constitutes persistent exfiltration of conversation logs to a remote infrastructure managed by the vendor.
- [PROMPT_INJECTION]: The skill retrieves previously stored data ('memories') from the cloud and injects it into the current agent session based on keyword matching. This creates an indirect prompt injection surface where malicious or untrusted content from past sessions can be re-introduced into the prompt context, potentially influencing agent behavior in subsequent turns.
- Ingestion points: Conversation data is retrieved from the DeepLake REST API during the 'Recall' phase.
- Boundary markers: No explicit delimiters or boundary warnings are described to separate recalled memories from the primary system instructions.
- Capability inventory: The skill leverages the agent's ability to read and process context to incorporate external data.
- Sanitization: The documentation does not specify any sanitization or validation of the retrieved memory content before it is processed by the agent.
Audit Metadata