webnn

Pass

Audited by Gen Agent Trust Hub on Aug 20, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute a local Node.js script (scripts/find-webnn-targets.mjs) to automate the discovery of frontend files and WebNN-related code patterns within the project directory.
  • [PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by reading and processing the contents of files in the workspace.
  • Ingestion points: The scripts/find-webnn-targets.mjs script reads file contents from the workspace using fs.readFileSync for analysis.
  • Boundary markers: The script lacks explicit boundary markers or instructions to ignore embedded commands within the files it processes.
  • Capability inventory: The script performs local file reads and logs findings to the console, with no observed network exfiltration capabilities.
  • Sanitization: No sanitization or filtering of the read file content is performed prior to processing the markers.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 20, 2026, 06:37 PM
Security Audit — agent-trust-hub — webnn