Model Context Protocol
Open standard for giving models tools and context.
- made by
- Anthropic
Magic Ship is one shop in Vancouver, BC, working remotely with clients worldwide. We are not a partner, reseller, or certified vendor of Anthropic - we just build with this.
What Model Context Protocol is
MCP is an open protocol for connecting models to tools, data, and prompts. It was introduced by Anthropic and is now developed in the open with implementations from several vendors. A server exposes tools, resources, and prompts over stdio or HTTP, and any compliant client can call them, which replaces one bespoke integration per client with one server many clients can use.
How we use it
When a client's internal system needs to be reachable by an agent, we build an MCP server for it once - typed tool schemas, real auth, error messages a model can act on - and it then works from Claude Code, from the product agent, and from whatever client comes next. Tools are scoped on purpose: read tools broad, write tools specific, destructive ones behind an approval step. On a forward-deployed engagement this is often the cleanest first deliverable.
Where it is the wrong choice
It is a plumbing standard, not an authorization model. Pointing an agent at a server that wraps a powerful internal API without scoping the tools and auditing the calls moves the risk rather than removing it, and for a single fixed integration a direct function call is less machinery to maintain.
Service lines it turns up in
Related tools
More in Agent frameworks
Other tools in the same service lines
Building something on Model Context Protocol?
Send the problem rather than a job spec. You get an answer on scope, on fit, and on whetherModel Context Protocol is even the right call for it.
Start a projectModel Context Protocol and Anthropic are trademarks of their respective owners, used here to say what we work with.