replit
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes the
replitCLI binary to perform actions. Supported commands includestatus,authorize-url,list-tools, andcall-tool. - [SAFE]: Implements OAuth 2.0 via a direct flow with
replit-mcp.comandreplit.com. The instructions explicitly forbid the agent from soliciting user credentials in chat, which mitigates phishing risks. - [SAFE]: Uses granular scopes (
apps:read,apps:write,offline_access) and specifiesdefault: askfor write-related methods (apps.create,apps.edit,apps.publish), ensuring human oversight for significant changes. - [SAFE]: The
list-toolscommand is restricted to reviewed tools, providing a layer of protection against unvetted third-party tool execution.
Audit Metadata