validation-doctor

Warn

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • EXTERNAL_DOWNLOADS (MEDIUM): The skill provides and encourages the use of npx to download and execute packages (chrome-devtools-mcp@latest and @brave/brave-search-mcp-server) directly from the npm registry.
  • Evidence: The configuration snippets use the -y flag with npx, bypassing confirmation prompts for package installation.
  • Risk: Use of the @latest tag instead of a specific version hash allows for potential code execution of compromised package versions if the upstream registry or account is breached.
  • COMMAND_EXECUTION (LOW): The skill involves executing shell commands (npx) and calling external tools (list_pages, brave_web_search).
  • Indirect Prompt Injection (LOW): The skill is designed to ingest and report on data from external, untrusted sources (Brave Search results and Chrome DevTools page content).
  • Ingestion points: Data returned from brave_web_search and list_pages / mcp6_list_pages (SKILL.md).
  • Boundary markers: Absent; there are no instructions to the agent to ignore malicious instructions embedded in the search results or page metadata.
  • Capability inventory: Command execution via npx and network access through the MCP tools themselves.
  • Sanitization: None detected in the skill definition.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 17, 2026, 06:12 PM