01Why this interface
Most MCP servers give the agent a long list of tools to call one at a time. For anything beyond a single lookup, the agent has to choose a tool, wait for the result to come back through the conversation, reason about it, and choose again, and that loop repeats at every step. It is slow, and the agent often takes a wrong turn before it reaches the answer.
Armature takes the approach known as code mode. Instead of a menu of tools, the agent gets a typed API to program against: it writes one short TypeScript task that carries out the whole job and runs it in a secure sandbox, which only calls back into your MCP for the operations you choose to expose.
The agent works through a single interface for discovery and execution, instead of choosing between dozens of separate tools.
Lookups, filters, joins, and retries run inside one program, so the agent only returns to the model once it has a finished result.
Each task runs in a sandbox built for fast startup and low memory, so runs stay short and cheap.
Product credentials live in server-side bindings, never in the task code an agent writes.
One task program routes through Armature.
Secure sandboxes
02Inputs
The lowest-friction path is the Armature GitHub App. Install it once, choose the repository and OpenAPI path, and interface builds become automatic from pull request to production. Armature starts from feature parity with the API, then lets you remove capabilities from the agent-facing surface before they ship.
03Generated surface
From the OpenAPI document, Armature generates an operation map, a typed virtual SDK, searchable documentation, and an agent-native search layer tuned to find the right operations, examples, and allowed capabilities quickly. The generated interface can match the API at launch, then exclude operations, resources, or whole capability groups that should stay out of the agent path.
// Generated SDK shape
const account = await product.accounts.getAccount({ accountId });
const tickets = await product.support.listTickets({ accountId, status: "open" });
return { account, openTicketCount: tickets.length };The generated docs explain namespaces, methods, arguments, response shapes, examples, timestamps, pagination, and auth assumptions. The runtime preserves the API's real errors, so agents learn from the behavior they will see in production.
04GitHub app flow
Install the Armature GitHub App, point it at an OpenAPI file, and every PR gets an automated interface preview: parse checks, generation checks, and a surface diff before the hosted MCP updates.
05Why Armature builds it
Armature builds and hosts the interface so your team gets the most performant agent-facing interface possible. You get a production-ready agent endpoint without owning the runtime work. It includes fast execution, separate preview and production environments, GitHub automation, and Armature reporting from day one.
Task programs can complete multi-step work inside the secure sandbox, then return a concise result.
GitHub automation keeps interface changes reviewable before they reach the hosted MCP.