MuleSoft has published the methodology behind its MCP cost claims, which is rarer than the claims. In MCP Cost Management: Inside Our Benchmarks, the team explains how it measured five gateway policies that shrink what MCP servers send to agents: progressive disclosure, smart response trimming, tool mapping, compressing repeated structures and cleaning payloads. It ran them against 18 public MCP servers, from Notion and Redis to Jira, GitHub, Stripe and arXiv, through its Omni Gateway, and compared each against a no-policy baseline.
The figures are MuleSoft’s own, and the caveats are the best part of the post. The savings vary enormously by server. Progressive disclosure, which hides tools a task does not need, cut 91.8 percent of MCP tokens on Notion and 87.6 percent on Redis, and nothing on arXiv; smart response trimming cut 93.9 percent on arXiv and 60.4 percent on Stripe. Apify, whose own measurements this series covered two weeks ago, saw 56.7 percent from progressive disclosure. The post is explicit that the percentages “cannot be directly added together,” that they measure only MCP tokens at the gateway, not the agent’s own prompts and reasoning, and that they “do not establish equivalent reductions in a complete agent workflow or prove unchanged answer quality.” The method story is the lesson: a first benchmark against live servers failed because agents took different numbers of turns, and “a run with one extra turn can consume substantially more tokens even when the policy reduces the individual payloads.” So the team froze responses into a golden set and judged answer quality separately.
The catalog holds the platform this runs on. The MuleSoft provider page lists API pages including the MuleSoft Anypoint Platform and the MuleSoft Anypoint Exchange API, where the MCP servers a gateway governs are catalogued. On the Agent Readiness score, a well-known catalog, an agent card, delegated identity and protected resource metadata are lit, a rarer set than most of the catalog has.
The Kin Score is 48.3, developing band, with contract quality at 65.2 and developer ergonomics at 65.5 against access clarity at 25.0. The Agent Readiness score is 43.4. The MCP server dimension is still unlit on the record of a company publishing benchmarks of other people’s MCP servers, as it was when this series covered MuleSoft’s open policies last week. The post’s honesty is worth copying: it names exactly what the measurement covers and what it does not. The rule it arrives at, that turn count dominates per-call savings, is the same one TestMu AI found from Slack’s data this week. Two vendors, two methods, one finding: an agent’s cost is the task, not the call.