diffbot
Pass
Audited by Gen Agent Trust Hub on Apr 22, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill documentation instructs the installation of the
@membranehq/cliglobal package via npm. This is a vendor-owned utility used to manage the authentication lifecycle and facilitate communication with the Diffbot API. - [COMMAND_EXECUTION]: The skill utilizes shell-based interactions with the
membraneCLI to perform various tasks, includingmembrane login,membrane search,membrane connect, andmembrane action run. These commands are used to manage authentication and trigger data extraction workflows. - [PROMPT_INJECTION]: The skill has an inherent surface for indirect prompt injection because its core function is to extract and process data from external, untrusted web pages.
- Ingestion points: Data is ingested into the agent's context from external URLs via Diffbot's extraction tools (Articles, Products, Discussions, etc.).
- Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions to separate extracted data from agent instructions.
- Capability inventory: The skill allows the execution of arbitrary API requests via
membrane requestand pre-defined actions viamembrane action run. - Sanitization: No sanitization or validation logic is specified for the content extracted from third-party web pages before it is processed by the agent.
Audit Metadata