DeepWiki MCP server tool surface audit

A100/100composite score
Token footprint
~356
Tools
3
Findings
0
Scanned

Engine mcp-surface-lint@0.3.3 · reported as DeepWiki v2.14.3

The DeepWiki MCP server exposes 3 tools whose definitions cost approximately 356 tokens in every single conversation, whether or not any of them is ever called. MCP Surface Lint scores that surface 100 out of 100 (grade A) with no scored findings against it.

Its strongest category is surface at 100/100; the weakest is annotations at 100/100. It is catalogued here under Developer Tools, alongside other servers agents reach for in the same kind of work.

Token footprint

Measured against a 200,000-token window, the surface consumes about 0% of it before the conversation starts, and keeps consuming it on every turn. That averages 119 tokens per tool definition, counted with the o200k_base encoding.

Compared with the Developer Tools median of 2,797 tokens, DeepWiki is roughly 87% below.

surface100
naming100
descriptions100
schemas100
annotations100
design100

Findings

Nothing in the catalogue fires on this surface. DeepWiki passes all 19 rules: 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, the input schemas are typed tightly enough to constrain what a model sends, read and write operations are machine-distinguishable before a client runs them and the surface is shaped around user intents rather than mirrored REST endpoints.

A clean result is not a claim that the server is well designed for your particular workflow — only that its published tool surface holds up against every static check we run.

Tool inventory

The heaviest definition is `ask_question` at 147 tokens, roughly 41% of the total; the lightest, `read_wiki_contents`, costs 103. Every tool carries a description, ranging from 45 to 92 characters.

None declare annotations, so a client cannot tell a read from a write without reading the prose. Across the surface, schemas carry 1 top-level parameters on average and reach a maximum depth of 2.

All 3 tools as published, heaviest first. Token counts use the o200k_base encoding and are approximate.

Every tool on this surface with its token cost and schema shape.
ToolTokensDescriptionParamsRequiredDepthAnnotations
ask_question14792 chars222no
read_wiki_structure10759 chars111no
read_wiki_contents10345 chars111no

How this compares

Within Developer Tools this is rank 1 of 5 scanned servers — 4 points above the category median of 96.

The closest comparisons in this directory are Context7, Filesystem and Playwright, which sit nearest on score and face the same kind of surface-design pressure.

Category
Developer Tools
Rank in category
1
Category median score
96
Category median tokens
2,797
Want this surface fixed, not just measured?

A fix plan turns these findings into an ordered set of changes with the token saving of each one.

Get a fix plan

This page is a static analysis of DeepWiki'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.