Skills
skills/moeru-ai/airi/agent-browser-electron/Gen Agent Trust Hub

agent-browser-electron

Pass

Audited by Gen Agent Trust Hub on May 6, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [SAFE]: No malicious patterns, exfiltration attempts, or obfuscated code were detected in the skill instructions or metadata.
  • [COMMAND_EXECUTION]: The skill instructs the agent to use curl and agent-browser to communicate with the local CDP server at 127.0.0.1. These operations are standard for Electron debugging and are confined to the local loopback interface.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface. 1. Ingestion points: agent-browser snapshot, agent-browser console, and agent-browser errors in SKILL.md. 2. Boundary markers: Absent. 3. Capability inventory: Browser automation (navigation, interaction) via the agent-browser tool. 4. Sanitization: Absent. This surface is an inherent part of browser automation and debugging, and given the local context and lack of host-level execution capabilities for the ingested data, it is considered safe.
Audit Metadata
Risk Level
SAFE
Analyzed
May 6, 2026, 04:21 PM