next-dev-loop
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- External Package Management: The skill suggests using standard package managers to install or upgrade the
nextframework and theagent-browsertool. These are well-known resources in the development ecosystem and are necessary to meet the skill's version requirements. - Command Execution: The skill involves executing shell commands to drive a browser automation tool and probe internal framework endpoints. These operations are core to the skill's purpose of verifying runtime behavior.
- Browser Session Persistence: The workflow uses session identifiers to maintain browser state, such as cookies and local storage, across commands. This is a common practice to facilitate testing in authenticated environments without requiring repeated manual login steps.
- Indirect Prompt Injection Surface: The skill ingests data from a running application, including DOM content and console logs. As with any tool that processes external web content, there is a theoretical surface for indirect prompt injection if the application under test renders untrusted data, though this is consistent with the skill's intended use case.
Audit Metadata