web-js-reverse-master-flow
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data, including third-party website content and JavaScript source code.
- Ingestion points: The agent captures live site data, DOM structures, and JS snippets through tools like
chrome-devtools-mcpandjs-reversebased on user-provided URLs. - Boundary markers: The instructions do not specify the use of delimiters or clear separation between the agent's instructions and the untrusted data being analyzed, which could allow a malicious script to influence the agent's behavior.
- Capability inventory: The workflow involves writing detailed analysis records to the local file system (e.g.,
reverse-records/) and executing potentially dangerous code in browser or Node.js environments during the emulation phases. - Sanitization: There are no explicit requirements for the agent to sanitize or validate the captured content before interpreting it, creating a surface for adversarial content to affect the analysis logic.
Audit Metadata