Where does MCP server token usage come from?
Tool definitions occupy context. Tool outputs often return far more data than the next step needs. Multi-step chains send intermediate results back through the model, and every new run may reconstruct the same plan. Progressive disclosure helps schema overhead, but it does not eliminate repeated reasoning after a workflow stabilizes.
What are the main MCP token optimization techniques?
In rough order of effort: connect only the servers a task needs, since every connected server's schemas occupy context whether used or not. Prefer servers that support progressive disclosure, loading tool definitions on demand. Filter and truncate tool results in code before they reach the model. Let the agent execute code against tool outputs instead of reading them. And for procedures that have stabilized, compile the whole multi-tool sequence into a workflow so repeated runs stop paying for orchestration at all.
Code execution validates the direction
Anthropic's engineering team showed a 150,000-to-2,000-token example by letting an agent write code that filtered and joined MCP results outside the context window. Rote takes the next step for recurring work: validate the discovered procedure, version it, and run it again without regenerating orchestration code every session.
Keep MCP as the integration layer
Rote's MCP backend can retain authenticated MCP external nodes for DBOS, DBOS TypeScript, Inngest, and Cloudflare Workers. Direct API backends are also available as explicit implementation boundaries. This is a composition strategy, not a replacement protocol.
Expose the result as one tool
rote serve can present deployed compiled pipelines to MCP clients as typed tools for triggering runs, checking status, and sending durable signals. Agents interact with the business capability rather than carrying its internal orchestration through the model context.
Frequently asked questions
Why are MCP workflows expensive in tokens?
The main cost comes from agent context and orchestration: tool schemas, large intermediate results, and repeated model decisions between calls. The protocol is not the root problem; repeatedly routing a stable procedure through a model is.
Does Rote replace MCP?
No. A compiled workflow can call authenticated MCP servers, or use direct API integrations. Rote can also expose the deployed workflow as one typed MCP tool, so an agent triggers a durable workflow instead of replaying the original multi-step skill.
Can MCP tool calls run without every result passing through the model?
Yes. Code inside a compiled workflow can call tools, filter results, and pass only the data required by a retained judgment node. Deterministic nodes do not need to round-trip through model context.
How do I measure MCP token consumption?
Compare input token counts for the same task with and without each server connected; the difference is that server's schema overhead. Then inspect a full transcript and attribute tokens to tool definitions, tool results, and the model's reasoning between calls. The last two usually dominate, which is why filtering results and compiling stable procedures outweigh trimming schemas.
Compile a workflow
Inspect the open-source CLI or run a compilation in Rote Cloud.