Features

MCP Servers

Add Model Context Protocol servers with a single paste. Full health monitoring and OAuth support.

What is MCP?

The Model Context Protocol (MCP) is an open standard that lets AI assistants connect to external tools and data sources. Think of MCP servers as plugins — they give Claude access to databases, APIs, documentation, and more.

Conare makes MCP management visual and painless.

Adding a Server

One-Paste Setup

Most MCP providers give you a command like:

claude mcp add context7 -- npx -y @context7/mcp

In Conare, open the MCP Panel, click Add Server, and paste that command directly. Conare parses it automatically:

  • Server name → context7
  • Transport → stdio (default)
  • Command → npx
  • Args → -y @context7/mcp
  • Flags → --env, --header, --scope all supported

Click Add and you're done. No JSON editing, no config files.

Manual Setup

For advanced use cases, the Add modal also has fields for:

  • Transport typestdio (local process), sse (Server-Sent Events), http
  • Environment variables — key-value pairs passed to the server process
  • Scopelocal (project only), project (workspace), user (global)

Server Health

Every server shows a real-time status indicator:

StatusIconMeaning
HealthyServer connected, tools available
Needs AuthOAuth required — click to authorize
ErrorConnection failed — check config
UnknownNot yet checked

Hover over any server to see detailed status information.

OAuth Flow

Some MCP servers require authentication (GitHub, Notion, etc.). Conare handles the full OAuth 2.1 PKCE flow:

Click the auth button

On a server showing "Needs Auth", click the key icon.

Authorize in your browser

Conare opens your browser to the OAuth provider's authorization page. A local callback server spins up automatically.

Token saved automatically

After you approve, the token is captured and injected into Claude Code's config. The server status flips to "Healthy".

Security

Token Cost Tracking

Each MCP server adds tools to Claude's context window, which costs tokens. Conare shows the exact token count per server — not just the number of tools, but the actual context window impact.

This appears in the server details modal alongside:

  • List of available tools with descriptions
  • Server transport type and config
  • Enable/disable toggle

Managing Servers

  • Enable/Disable — Checkbox toggle per server. Disabled servers use zero tokens.
  • Delete — Hover to reveal the trash button. Confirmation modal prevents accidents.
  • Scope visibility — Servers show their scope (local/project/user) so you know what's project-specific vs global.

Supported Transports

TransportUse CaseExample
stdioLocal CLI tools, npm packagesnpx @context7/mcp
SSERemote servers with streaminghttps://mcp.example.com/sse
HTTPREST-based MCP endpointshttps://api.example.com/mcp

Most community MCP servers use stdio transport. Remote servers typically use SSE or HTTP.

Here are some commonly used MCP servers that work great with Conare:

  • Context7 — Up-to-date docs for any library
  • Filesystem — Read/write files outside the project
  • GitHub — Issues, PRs, repos
  • Postgres/SQLite — Direct database access
  • Brave Search — Web search from Claude
Copyright © 2026