hexagone-web-feature-extractor
Pass
Audited by Gen Agent Trust Hub on Mar 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill orchestrates the execution of local Node.js scripts (
screenshot-server.jsandgenerate-docx.js) to manage screenshot data transfers and document generation. These operations are required for the skill's primary purpose. - [PROMPT_INJECTION]: The skill ingests untrusted content from targeted web pages to extract functional descriptions and metadata. This creates a surface for indirect prompt injection where malicious instructions embedded in the web content could influence the agent's behavior during report generation.
- Ingestion points: Data extracted from Hexagone Web pages using JavaScript and browser automation (
SKILL.md). - Boundary markers: No explicit delimiters or boundary instructions are utilized when processing extracted text.
- Capability inventory: File system write access for screenshots and documents, local network listener (port 8765), and subprocess execution for script management.
- Sanitization: No validation or sanitization of the extracted web content is implemented before its inclusion in the metadata or final output documents.
- [DATA_EXFILTRATION]: The skill interacts with internal vendor-specific network locations (
dedalus.lan) and local host addresses. No evidence of data transmission to external or untrusted domains was found.
Audit Metadata