Security audit
Security checks for vulnerabilities and agentic risk
The skill is a disclosed, read-only Compass real-estate lookup integration, but users should understand it routes requests through their signed-in Compass browser session.
Install only if you are comfortable letting an MCP server and browser extension fetch Compass pages through your signed-in Compass browser tab. Treat the npm package and extension as trusted components, and avoid using it with Compass accounts or browser profiles that contain data you do not want exposed to the agent workflow.
Skill accesses MCP server configuration files (mcp.json). MCP configs contain server URLs, authentication tokens, and tool definitions — reading them allows the skill to discover and potentially abuse other tool integrations.
.mcp.json (project) or ~/.claude/mcp.json (global):
{
The trigger text is unusually broad because it activates on generic phrases like 'what does compass say about' and broadly on 'any request involving Compass properties, prices, or comparisons.' That can cause over-invocation of the skill and silently route user queries into a signed-in browser-backed data path, increasing the chance of unintended access or disclosure through the user's authenticated Compass session.
The skill description explains the browser-routing mechanism, but it does not present a clear privacy warning that requests are sent through the user's signed-in Compass tab and may use account-scoped session context. Because the tool depends on an authenticated browser session, users may unknowingly expose account-linked browsing context or retrieve data under their identity without informed consent.
No suspicious patterns detected.