id-dataweb
Pass
Audited by Gen Agent Trust Hub on Apr 29, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill requires the installation of the
@membranehq/clipackage from the npm registry to facilitate communication with the Membrane platform. - [COMMAND_EXECUTION]: The instructions involve executing shell commands such as
membrane loginandmembrane action runto manage connections and process identity verification data. - [PROMPT_INJECTION]: There is a potential for indirect prompt injection as the agent ingests data from external ID DataWeb actions. * Ingestion points: Data entering the agent context via the output of the
membrane action runcommand in SKILL.md. * Boundary markers: There are no explicit delimiters or instructions provided to isolate or ignore instructions embedded within the external data. * Capability inventory: The agent has capabilities to execute shell commands and platform actions as described in SKILL.md. * Sanitization: No sanitization or validation of the external content is implemented before the agent processes it.
Audit Metadata