Every conversation with the Cloudflare Docs MCP server begins with 2 tool definitions already in context — about 421 tokens of it. Against the 19 static rules in the catalogue it scores 99 out of 100 — grade A — raising 3 scored findings in 2 of the six categories.
Surface is the strongest category at 100/100, and schemas the weakest at 95/100. It is catalogued here under Cloud and Infrastructure, alongside other servers agents reach for in the same kind of work.
At 421 tokens the surface takes about 0% of a 200,000-token context window, and it takes it on every turn — this is rent, not a one-off cost. That averages 211 tokens per tool definition, counted with the o200k_base encoding.
That lands exactly on the Cloud and Infrastructure median of 421 tokens.
The findings below are grouped by severity, and each links to the rule that produced it.
`design/list-no-limit` accounts for one finding. A list/search tool with no limit or pagination parameter can dump an unbounded payload into the context window. The caller must be able to ask for less. On this surface it fires on `search_cloudflare_documentation`.
One finding comes from `schemas/loose`. Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes. On this surface it fires on `migrate_pages_to_workers_guide`.
One finding comes from `schemas/param-desc-missing`. Undescribed parameters get filled by guesswork. Every property the model must populate deserves a sentence saying what it means and what shape it takes. On this surface it fires on `search_cloudflare_documentation`.
Findings that measurably degrade tool selection.
A list/search tool with no limit or pagination parameter can dump an unbounded payload into the context window. The caller must be able to ask for less.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Undescribed parameters get filled by guesswork. Every property the model must populate deserves a sentence saying what it means and what shape it takes.
4 of the six categories come through clean: surface, naming, descriptions and annotations. In practice that means the surface stays inside a sensible tool and token budget, the tool names follow one convention end to end, every tool carries a description long enough to disambiguate it and read and write operations are machine-distinguishable before a client runs them.
`search_cloudflare_documentation` is the most expensive definition at 350 tokens — about 84% of the whole surface — against 69 for `migrate_pages_to_workers_guide`. Every tool carries a description, ranging from 66 to 538 characters.
All 2 declare annotations, so a client can tell a read from a write before running one. Input schemas average 1 top-level parameters and nest 1 level deep at their worst, with 1 tool taking no input at all.
All 2 tools as published, heaviest first. Token counts use the o200k_base encoding and are approximate.
| Tool | Tokens | Description | Params | Required | Depth | Annotations |
|---|---|---|---|---|---|---|
search_cloudflare_documentation | 350 | 538 chars | 1 | 1 | 1 | yes |
migrate_pages_to_workers_guide | 69 | 66 chars | 0 | 0 | 0 | yes |
Cloudflare Docs ranks 1 of 1 scanned Cloud and Infrastructure servers, exactly on the category median of 99.
The closest comparisons in this directory are Context7, PostgreSQL and Slack, which sit nearest on score and face the same kind of surface-design pressure.
A fix plan turns these findings into an ordered set of changes with the token saving of each one.
This page is a static analysis of Cloudflare Docs's published tools/list surface on the date shown. It is not a security review, a performance benchmark, or a judgement of the product behind the surface. Read how the score is computed, or browse every audited server.