The Chrome DevTools MCP server exposes 29 tools whose definitions cost approximately 5,630 tokens in every single conversation, whether or not any of them is ever called. The composite score is 95 out of 100 (grade A), built from 32 scored findings touching 3 of the six rule categories.
The spread runs from 100/100 for naming down to 85/100 for surface. It is catalogued here under Developer Tools, alongside other servers agents reach for in the same kind of work.
Measured against a 200,000-token window, the surface consumes about 3% of it before the conversation starts, and keeps consuming it on every turn. That averages 194 tokens per tool definition, counted with the o200k_base encoding.
Against a category median of 2,797 tokens, this surface runs about 101% above the middle of the pack.
The findings below are grouped by severity, and each links to the rule that produced it.
30 findings come 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 `click`, `close_page` and `drag`, among 26 others.
`design/list-no-limit` contributes 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 `list_pages`.
`surface/tool-budget` accounts for one finding. Tool selection accuracy degrades as the surface grows: more near-neighbours to confuse, more tokens spent before the first user message. Fewer, intent-shaped tools outperform one-tool-per-endpoint surfaces. It applies to the surface as a whole rather than to one tool.
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.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Loose schemas (bare objects, untyped arrays, open additionalProperties) turn schema validation off exactly where the model is most likely to hallucinate argument shapes.
Tool selection accuracy degrades as the surface grows: more near-neighbours to confuse, more tokens spent before the first user message. Fewer, intent-shaped tools outperform one-tool-per-endpoint surfaces.
Advisory only — these deduct nothing from the score.
readOnlyHint/destructiveHint are machine-readable: clients use them to decide when to ask the user for confirmation. Prose like 'confirm with the user first' is invisible to that machinery.
When the same enum is reused across tools but not every combination is valid (e.g. not every product supports every tier), the schema cannot say so — the model finds out via runtime errors. Flag shared enums for a validity review.
handle_dialog/properties/action; navigate_page/properties/handleBeforeUnload
3 of the six categories come through clean: naming, descriptions and annotations. In practice that means 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.
`emulate` is the most expensive definition at 413 tokens — about 7% of the whole surface — against 66 for `list_pages`. Every tool carries a description, ranging from 30 to 318 characters.
All 29 declare annotations, so a client can tell a read from a write before running one. Input schemas average 4 top-level parameters and nest 3 levels deep at their worst, with 1 tool taking no input at all.
All 29 tools as published, heaviest first. Token counts use the o200k_base encoding and are approximate.
| Tool | Tokens | Description | Params | Required | Depth | Annotations |
|---|---|---|---|---|---|---|
emulate | 413 | 45 chars | 8 | 1 | 1 | yes |
list_console_messages | 328 | 72 chars | 7 | 1 | 2 | yes |
evaluate_script | 315 | 133 chars | 6 | 2 | 2 | yes |
list_network_requests | 273 | 77 chars | 5 | 1 | 2 | yes |
take_screenshot | 273 | 41 chars | 6 | 1 | 1 | yes |
navigate_page | 262 | 85 chars | 7 | 1 | 1 | yes |
fill_form | 261 | 318 chars | 3 | 2 | 3 | yes |
performance_start_trace | 245 | 151 chars | 4 | 1 | 1 | yes |
get_network_request | 235 | 246 chars | 4 | 1 | 1 | yes |
new_page | 214 | 74 chars | 4 | 1 | 1 | yes |
take_snapshot | 206 | 302 chars | 3 | 1 | 1 | yes |
upload_file | 193 | 41 chars | 4 | 3 | 2 | yes |
lighthouse_audit | 192 | 173 chars | 4 | 1 | 1 | yes |
press_key | 186 | 166 chars | 3 | 2 | 1 | yes |
performance_analyze_insight | 184 | 144 chars | 3 | 3 | 1 | yes |
fill | 182 | 79 chars | 4 | 3 | 1 | yes |
wait_for | 157 | 59 chars | 3 | 2 | 2 | yes |
click | 152 | 30 chars | 4 | 2 | 1 | yes |
drag | 150 | 36 chars | 4 | 3 | 1 | yes |
performance_stop_trace | 139 | 66 chars | 2 | 1 | 1 | yes |
handle_dialog | 137 | 61 chars | 3 | 2 | 1 | yes |
take_heapsnapshot | 133 | 128 chars | 2 | 2 | 1 | yes |
type_text | 133 | 56 chars | 3 | 2 | 1 | yes |
hover | 131 | 31 chars | 3 | 2 | 1 | yes |
get_console_message | 126 | 92 chars | 2 | 2 | 1 | yes |
select_page | 121 | 49 chars | 2 | 1 | 1 | yes |
resize_page | 119 | 66 chars | 3 | 3 | 1 | yes |
close_page | 102 | 66 chars | 1 | 1 | 1 | yes |
list_pages | 66 | 40 chars | 0 | 0 | 0 | yes |
Chrome DevTools ranks 5 of 5 scanned Developer Tools servers, 1 point below the category median of 96.
The closest comparisons in this directory are Filesystem, Playwright and Context7, 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 Chrome DevTools'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.