ui-primitive-picker
Pass
Audited by Gen Agent Trust Hub on Aug 7, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill ingests untrusted local data when inspecting project lockfiles and source code to establish the current stack, creating a surface for indirect prompt injection. 1. Ingestion points: SKILL.md (Inspect the project first section). 2. Boundary markers: Absent. 3. Capability inventory: File system read access and command execution (package manager). 4. Sanitization: Absent.
- [COMMAND_EXECUTION]: The skill instructs the agent to perform dependency installation using the project's package manager. This action is mitigated by a required human-in-the-loop approval step ('Ask for approval before adding a dependency').
- [EXTERNAL_DOWNLOADS]: The agent is directed to access external documentation and repositories to verify the maintenance status and features of UI libraries, which involves network access to third-party domains.
Audit Metadata