glossary

what is mcp (model context protocol)?

MCP is a standard that lets AI agents connect to external tools. GitHub, Linear, Jira, Slack, databases, docs servers, anything. Instead of each agent building its own integration for each tool, MCP gives them one shared way to do it.

the problem mcp solves

Without MCP, if you want Claude Code to read your Linear tickets, you'd need someone to build a Claude-specific Linear integration. Want Codex to do the same thing? Someone builds a Codex-specific integration. Every agent, every tool, every combination needs its own glue code.

MCP fixes this by creating one standard. A tool implements MCP once, and every agent that speaks MCP can use it. An agent implements MCP once, and it can use every MCP-compatible tool.

how it works in practice

You create a .mcp.json file in your project:

{
  "mcpServers": {
    "context7": {
      "command": "npx",
      "args": ["-y", "@upstash/context7-mcp@latest"]
    }
  }
}

When your agent starts, it reads this file and connects to Context7 automatically. Now it can look up library documentation on the fly. Add more entries to connect to more tools.

mcp in pane

Pane doesn't do anything special with MCP. A Pane terminal is a real terminal. When Claude Code runs inside it, it reads .mcp.json the same way it would in any standalone terminal.

The nice part: since each pane runs in its own folder, each pane can have its own MCP config. One pane connects to your company's internal tools, another connects to public docs. No conflicts.

why pane uses a cli instead of mcp

MCP is great for connecting agents to external tools. But for Pane's own workspace control (creating panes, reading terminal output, sending input), a CLI makes more sense. Every agent already knows how to run terminal commands. No setup, no extra server, works everywhere.

further reading

by Parsa Khazaeepoul

Last updated June 24, 2026