MCP servers speak a common protocol, so the same server you configured elsewhere often drops straight into Crush’s config with a path change.

What is MCP and why does it matter here?

MCP is an open protocol that lets an AI client discover and call tools exposed by a server process. The client asks “what can you do,” the server replies with a list of functions and their schemas.

For a coding agent, that means tooling like a structured issue-tracker client, a schema-aware database query tool, or a doc search — anything with a maintained MCP server.

How do you register an MCP server?

1. Locate Crush’s configuration file, typically a JSON file in its config directory.
2. Add an mcp entry with a command that starts the server — an npx package or a binary path.
3. Provide any required environment variables, such as an API token for the service.
4. Restart Crush so it loads the new server.
5. In a session, ask a question that needs the tool and confirm it appears.

Most servers are launched as a subprocess over stdio, so a typo in the command means Crush silently has no tools from it. Check the startup output when troubleshooting.

What makes a good MCP tool to add?

How do you scope permissions?

Give each server the least access it needs. A database tool should connect with a read-only role unless you specifically need writes. An API token should be scoped to the one service, not a full account key.

Crush still asks for confirmation before running tools, which is a real safety layer. Keep it on for any tool that mutates state — the model can misread a request and call a write tool when you meant a read.

What goes wrong with MCP setups?

Version drift is common: a server package updates and changes its function signatures, so the model calls a tool with stale arguments. Pin versions. Startup failures are quiet, so verify the server responds before relying on it in a workflow.

Too many tools also degrade performance. Each tool’s schema consumes context, and dozens of options make the model pick the wrong one. Add servers deliberately, not in bulk.

Who should wire this up?

Developers who already run internal services and want the terminal agent to reach them. If your whole workflow is local files and git, the built-in file and shell tools cover it and MCP adds complexity for little gain. Add it when a real external system needs to be in reach.

Quick Summary

• Crush can call MCP servers, adding tools like database queries and doc search to a session.
• Register servers in Crush’s config with a start command and any env vars, then restart.
• Scope each server to least privilege and keep write tools behind confirmation.
• Pin server versions and avoid adding many tools — extra schemas consume context.

If you spend your day in a terminal and dislike switching to an editor panel to talk to an AI, this is built for you.

What is Crush?

Crush is a command-line coding assistant. It runs a session in your terminal, understands the project it sits in, and can read files, suggest edits, and run commands when you confirm them.

It supports multiple model providers, so you can point it at a hosted API or a local model depending on your budget and privacy needs.

Is Crush free?

The tool is free and open source. Model usage is not free unless you run a local model.

With a hosted key you pay per token through that provider. With a local model through a tool like Ollama, your only cost is hardware. There is no separate Crush subscription.

How do you install and start it?

1. Install via a package manager — brew install on macOS, or the provided script for Linux.
2. Run crush from inside your project directory so it picks up the repo context.
3. Set your provider and API key in its configuration file.
4. Type a request in plain language, such as “add input validation to the signup handler.”
5. Review each proposed file change and accept or reject it.

The first session may prompt for configuration. A few lines of JSON name your provider and model; after that it remembers.

What can it actually do?

How should you prompt it?

Be concrete about files and intent. “Refactor utils/date.ts to remove the deprecated format function and update callers” gets a cleaner result than “clean up the code.”

Work in small steps. One change per request keeps the diffs reviewable and makes it easy to spot when Crush misunderstands your intent.

Where does a terminal agent fall short?

No visual diff viewer — you read changes as terminal output, which is harder for large edits than a side-by-side pane. And it depends on your model’s context window; large files can get truncated, so split work when a file is huge.

It also asks for confirmation a lot, which is correct for safety but slows fast iteration. Adjust that only once you trust it in a given repo.

Who should use Crush?

Terminal-first developers who want AI help without leaving the shell, and anyone who values an open-source tool they can inspect. If you prefer a rich GUI with inline diffs and integrated previews, an editor-based assistant will feel more natural. Crush is for the command line, unapologetically.

Quick Summary

• Crush is a free, open-source terminal AI coding agent from Charm.
• Install via a package manager, configure a provider, and run it inside your repo.
• It reads files, proposes edits, and runs commands, all with your approval.
• It supports hosted and local models — you pay only for model usage, if any.