Where it fits
Teams already using Cortex Context who want its context source available in Claude Desktop or another stdio-capable client.
- Connect Claude Desktop to Cortex Context through the local adapter.
- Use a Cortex token and project ID to scope the connection.
- Run the same adapter in another MCP client that can start a local process.
A remote MCP server advertises tools to your client. The provider controls the tools, the data they can reach, and the permissions they need. Read the provider's docs before you use it with private data.
Skills that pair with Cortex Context
These skills cover adjacent tasks, matched by purpose and the work each one describes.
Dispatches one agent per independent problem domain so unrelated failures can be investigated concurrently, then reviews and integrates the results. Use it when failures are separate, need no shared context, and can run without interfering.
For carrying out an implementation plan inside the current session, with the agent acting as the implementer itself. That happens when the human partner picked inline execution or when no subagent tool is available.
Helps decide how to integrate finished work after implementation and passing tests. Presents structured choices for merging locally, pushing and creating a pull request, or keeping the branch.
No skills found
Try a different search term or clear the filters.
Run the Cortex Context adapter
The adapter communicates over stdio and is started by the client. The hosted vercel URL is a 401 connectivity probe, not a documented MCP transport. Register at https://mcp.cortexguardai.com/ for credentials.
npx @cortexguardai/mcp@latest --token <TOKEN> --project-id <PROJECT_ID>This is a local install path. Install and run it first, then add the local command or URL to OpenCode, VS Code, or another compatible client. The provider controls the package, required keys, and runtime.
Common failure modes
These are the problems to check first. If the provider changes its endpoint or authentication flow, use its documentation as the source of truth.
- Authentication returns 401 or 403. Check the key, its environment variable, and the permissions attached to it.
- The key works in the provider dashboard but not in the client. Confirm the client sends it in the header the provider expects.
- The client cannot find the vomo command. Confirm the global CLI install and that the executable is on PATH.
- The process starts but no tools appear. Sign in to the provider CLI, restart the client, and check the local command output for errors.
Before you connect
- The MCP server process runs on your machine, not on a provider host.
- Data leaves only through the model provider and any MCP servers you attach.
- Review the local command, package, and files it can reach before running it.
- Stop the process or remove the client entry when you no longer need it.
Check the connection before a real task
These guides cover the decisions that a provider description cannot answer: endpoint ownership, scopes, tool access, and a read-only test.
Provider source
Setup requirements, available tools, and rate limits live with the provider. Use the source link as the authority for this server.
Cortex Context repositoryhttps://github.com/cortexguardai/mcp