web-development
Pass
Audited by Gen Agent Trust Hub on Nov 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill uses the
agent-browsertool to validate web applications, which involves reading content from web pages. This establishes an attack surface where an external web page could potentially contain malicious instructions aimed at the agent. - Ingestion points: Browser-based validation flows documented in
browser-testing.mdandSKILL.md. - Boundary markers: The skill does not explicitly define delimiters for untrusted web content, though it frames the browser's role strictly for verifying engineering outcomes.
- Capability inventory: The skill has access to file writes (code implementation), command execution (npm build/test), and network access (via browser tool).
- Sanitization: No specific sanitization or filtering of browser-read content is mentioned.
- [SAFE]: The skill references official Tencent CloudBase resources including SDKs (
@cloudbase/js-sdk,@cloudbase/node-sdk) and documentation domains (cloudbase.net). These are legitimate vendor resources consistent with the skill's stated purpose.
Audit Metadata