openclaw-control-ui-e2e
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute local repository scripts and package manager commands such as
node scripts/run-vitest.mjsandpnpm test:ui:e2eto perform end-to-end tests. These commands allow the agent to run code within the local environment. - [INDIRECT_PROMPT_INJECTION]: The skill implements a UI Stress Test workflow that generates HTML galleries containing feedback inputs. This data is persisted locally and exported back into the agent conversation, creating a surface for indirect prompt injection.
- Ingestion points: Feedback inputs within the generated UI stress-test gallery and the resulting exported text (referenced in
references/upstream/SKILL.source.md). - Boundary markers: The instructions emphasize separating mocked-Gateway evidence from real product data and require explicit disclosure review before uploading recordings.
- Capability inventory: Local command execution (
node,pnpm) and browser automation tools (playwright) capable of screenshots and video recording. - Sanitization: There is no documented sanitization or filtering for the user-provided feedback data before it is re-ingested into the agent's context.
Audit Metadata