budibase
Pass
Audited by Gen Agent Trust Hub on Apr 28, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill requires the installation of the
@membranehq/cliNPM package. This is a trusted vendor tool used for platform interactions. - [COMMAND_EXECUTION]: Uses shell commands through the
membraneutility for managing connections and executing actions. These are legitimate operations for the skill's purpose. - [DATA_EXFILTRATION]: Employs secure best practices by using Membrane's server-side connection management, ensuring that sensitive credentials like API keys are not handled or stored locally by the agent.
- [PROMPT_INJECTION]: The skill ingests untrusted data from Budibase records. 1. Ingestion points: Budibase app data, tables, and rows (SKILL.md). 2. Boundary markers: Absent. 3. Capability inventory: CLI access to create and run actions via
membrane action run(SKILL.md). 4. Sanitization: Not specified in the current skill documentation.
Audit Metadata