by Trung Thanh Tran

Turbo Browser MCP review, pricing and limits

Batched Playwright browser control for agents, with a default submit guard

Best for
AI agent developer
Not for
Non-technical user who wants a point-and-click tool
  • Browser control
  • Automation
  • Developer

Should you use Turbo Browser MCP?

Pick it if your coding agent runs long browser tasks or fills forms and you want fewer tool calls plus a default stop before the final click. It is the wrong choice for hosted browsers, stealth or Lighthouse audits. Today it needs Node 20 and a git clone, and the project is only days old.

One call can run a whole multi-step flow and return only the elements that changed at the end.

Also for

  • Automation engineer
  • Developers building browser-driving agents in Claude Code, Cursor or Codex
Show 2 more
  • Engineers automating long web forms and portals where a person must approve the last click
  • Teams that pay per token and run many multi-step browser sessions

Also not for

  • Team that needs a managed cloud browser

Main trade-off

A brand-new project with one maintainer, no npm package yet and benchmarks that nobody else has run.

Turbo Browser MCP pricing

MIT licence. Free to clone and run on your own machine. The repository lists no paid tier and no hosted service.

Install Turbo Browser MCP

git clone https://github.com/Tthanh2k6/turbo-browser-mcp.git && cd turbo-browser-mcp && npm install

Before you start

  • Chromium from npx playwright-core install chromium, unless Chrome, Edge or Brave is already installed
  • An MCP client entry that runs node on the absolute path of src/index.mjs, optionally with TURBO_BROWSER_SAVE_ROOT set
  • Chrome started with remote debugging on port 9222, only if you want the agent to reuse your open browser

How to invoke it

Exposed as MCP tools over stdio (tabs_batch, inspect_compact, smart_fill_form, act_fast and more) that the agent calls by name once the server is registered in its MCP client config.

Works with

  • Claude Code
  • Cursor
  • Antigravity
  • Codex
  • Any MCP client

Required runtime

  • Node.js 20 or newer (package.json engines)
  • playwright-core 1.49 or newer (package.json dependency)

Where it runs

  • Desktop
  • CLI

Data it can reach

  • web pages open in the browser
  • cookies and local or session storage
  • files inside the save root
  • console and network logs

The permissions this connector asks for, as its own documentation states them.

What Turbo Browser MCP can do

11 actions an agent can call. Open one for its parameters and an example.

tabs_batchRuns an ordered list of tab operations in one call: list, open, open_many, navigate, activate, resize, emulate, reload, back, forward, close and close_browser.

Example call

{"name":"tabs_batch","arguments":{"operations":[{"action":"open","url":"https://example.com/app","isolatedContext":"acct-a"},{"action":"open_many","urls":["https://example.com/docs","https://example.com/pricing"],"concurrency":4}]}}

Parameters

ParameterTypeNeededDescription
operationsarrayOptionalOperations executed in order. Each has an action plus fields such as url, urls, tabId, isolatedContext, width, height, colorScheme, userAgent or waitUntil.
operations[].concurrencynumberdefault 5OptionalTabs opened at once by open_many, from 1 to 10.
operations[].isolatedContextstringOptionalName of a separate context with its own cookies and storage, for the open action.

Returns

FieldTypeDescription
operations_resultobjectOutcome of each operation in order, with the tab ids that now exist.
inspect_compactReturns a size-capped snapshot of interactive elements, forms or readable sections across the main frame, iframes and open shadow DOM.

Example call

{"name":"inspect_compact","arguments":{"mode":"interactive","onlyChanged":true,"maxChars":3000}}

Parameters

ParameterTypeNeededDescription
onlyChangedbooleandefault falseOptionalLeave out elements unchanged since the previous snapshot.
modestringdefault interactiveOptionalOne of interactive, forms, article or all.
querystringOptionalKeep only elements matching this regex or substring.
maxCharsnumberdefault 4000OptionalCap on the snapshot length.
includeFramesbooleandefault trueOptionalLook inside child iframes.
boxesbooleandefault falseOptionalAdd bounding boxes to each snapshot line.
savePathstringOptionalFile path inside the working directory or save root where the full snapshot is written.

Returns

FieldTypeDescription
snapshotstringA header with tabId, character count, truncation flag and milliseconds, then one line per element with a short [eN] ref.
smart_fill_formFills a whole form in one call: text, selects, radios, checkboxes, sliders, rich editors, shadow DOM and iframe fields, a Gmail compose box and files from disk.

Example call

{"name":"smart_fill_form","arguments":{"fields":[{"selector":"#full_name","value":"Alex Nguyen"},{"selector":"#country","option":"United States"},{"selector":"#agree_terms","check":true},{"selector":"#resume_pdf","file":"./assets/resume.pdf"}],"stopBeforeSubmit":true,"returnSnapshot":true}}

Parameters

ParameterTypeNeededDescription
fieldsarrayOptionalFields to fill. Each targets a ref or selector and sets value, option, values, check, keys, file or files; frameIdx and all reach into iframes.
gmailobjectOptionalFills a Gmail compose dialog with to, subject and body.
overwritebooleandefault falseOptionalReplace values already present in a field.
stopBeforeSubmitbooleandefault trueOptionalStop before any submit action.
returnSnapshotbooleandefault falseOptionalAdd a form snapshot to the result.
snapshotMaxCharsnumberdefault 2500OptionalCap on the returned snapshot.

Returns

FieldTypeDescription
fill_reportobjectPer-field outcome with the tabId, and a form snapshot when returnSnapshot is true.
act_fastRuns one targeted page action: click, dblclick, hover, focus, clear, fill, type, select, check, uncheck, drag, drop, press, scroll, wait or upload.

Example call

{"name":"act_fast","arguments":{"action":"fill","selector":"#email","value":"alex@example.com","returnSnapshot":true}}

Parameters

ParameterTypeNeededDescription
actionstringRequiredOne of the 16 actions listed in the description.
refstringOptionalCompact element ref from a snapshot. selector, text, label or x and y can target instead.
valuestringOptionalText or option to enter for fill, type or select.
modifiersarrayOptionalKeys held during the action: Alt, Control, ControlOrMeta, Meta or Shift.
allowSubmitbooleandefault falseOptionalPermit an action that looks like Submit, Publish, Send or payment.
returnSnapshotbooleandefault falseOptionalAttach a compact snapshot to the response.
timeoutMsnumberOptionalTimeout for this action in milliseconds.

Returns

FieldTypeDescription
action_outcomeobjectThe action outcome. A blocked submit reports blockedBy as label, modal, network or ledger.
batch_actionsRuns an ordered list of page actions and form-fill steps in one call and stops safely if the submit guard blocks a step.

Example call

{"name":"batch_actions","arguments":{"steps":[{"action":"smart_fill_form","fields":[{"value":"Alex","selector":"#name"}]},{"text":"Next","action":"click"},{"action":"press","key":"Enter"}],"allowSubmit":true,"expect":{"textMatches":"Receipt\\s*R-\\d+","timeoutMs":8000},"submissionKey":"portal:2026-11-15:entry-1"}}

Parameters

ParameterTypeNeededDescription
stepsarrayRequiredOrdered steps: page actions or nested smart_fill_form steps. Must not be empty.
continueOnErrorbooleandefault falseOptionalKeep going when a step fails.
allowSubmitbooleandefault falseOptionalPermit steps that look like a final submit.
expectobjectOptionalProof to look for after an allowed submit: urlMatches, textMatches, selector and timeoutMs.
submissionKeystringOptionalIdempotency key written to a ledger; a repeat with the same key is refused.
allowDuplicatebooleandefault falseOptionalLet a repeated submissionKey through.
returnSnapshotbooleandefault falseOptionalAttach a snapshot taken after the last step.

Returns

FieldTypeDescription
step_outcomesobjectOutcome of every step. After an allowed submit it adds submitVerified and evidence: a new URL, matched text or a selector.
extract_clean_textExtracts readable plain text or compact markdown from the main or article content, without script and style noise.

Example call

{"name":"extract_clean_text","arguments":{"format":"markdown","maxChars":6000}}

Parameters

ParameterTypeNeededDescription
formatstringdefault textOptionaltext or markdown.
selectorstringOptionalLimit extraction to one element.
includeFramesbooleandefault trueOptionalInclude text from iframes.
maxCharsnumberdefault 6000OptionalCap on the returned text.
savePathstringOptionalWrite the full text to this file, which must sit inside the save root.

Returns

FieldTypeDescription
textstringA header with tabId, character count and truncation flag, then the extracted text.
evaluate_jsRuns a JavaScript expression or function in a frame or on one element, passing el and arg, and returns JSON-serializable output.

Example call

{"name":"evaluate_js","arguments":{"expression":"(el) => el.textContent","selector":"h1"}}

Parameters

ParameterTypeNeededDescription
expressionstringRequiredThe JavaScript expression or function to run.
frameIdxnumberdefault 0OptionalIndex of the frame to run in.
selectorstringOptionalElement to pass as el. ref or text can target instead.
argobjectOptionalAny JSON value passed to the function as arg.
savePathstringOptionalWhere to store the result; the path is confined to the save root.

Returns

FieldTypeDescription
valueobjectThe evaluated value as JSON with the tabId, cut at about 20,000 characters.
page_diagnosticsReads buffered console logs, network requests and failures, completed downloads and dialog logs for a tab, errors only by default.

Example call

{"name":"page_diagnostics","arguments":{"level":"errors","statusFilter":"errors"}}

Parameters

ParameterTypeNeededDescription
levelstringdefault errorsOptionalerrors or all console messages.
statusFilterstringdefault errorsOptionalerrors or all network requests.
includeStaticbooleandefault falseOptionalInclude 2xx images, fonts, stylesheets and media when statusFilter is all.
urlFilterstringOptionalKeep requests whose URL matches this substring or regex.
reqIdstringOptionalRequest id to inspect for headers, post data and response body.
partstringOptionalWith reqId, return only request-headers, request-body, response-headers or response-body.
dialogPolicyobjectOptionalHow to answer page dialogs: action dismiss or accept, with optional promptText.
clearbooleandefault falseOptionalClear the buffers after reading.

Returns

FieldTypeDescription
diagnosticsobjectConsole entries, network requests and failures, downloads and dialog events.
Show all 11 actions
capture_pageSaves a full-page or element screenshot as PNG or JPEG, or exports a PDF, inside the save root.

Example call

{"name":"capture_page","arguments":{"fullPage":true,"savePath":"./shots/home.png","mode":"screenshot"}}

Parameters

ParameterTypeNeededDescription
fullPagebooleandefault falseOptionalCapture the whole scrollable page.
modestringdefault screenshotOptionalscreenshot or pdf.
typestringdefault pngOptionalpng or jpeg.
qualitynumberdefault 80OptionalJPEG quality.
scalestringdefault cssOptionalcss or device pixel scale.
selectorstringOptionalCapture one element. ref or text can target instead.
savePathstringOptionalWhere the image or PDF is written, within the save root.
returnImagebooleanOptionalReturn the image inline as well.

Returns

FieldTypeDescription
fileobjectMetadata for the saved file, plus the image itself when returnImage is set.
network_sessionBlocks resources with a preset or custom patterns and reads, writes or clears cookies, storage and Playwright storageState files.

Example call

{"name":"network_session","arguments":{"preset":"turbo"}}

Parameters

ParameterTypeNeededDescription
presetstringOptionalturbo, no-images, no-media or clean (clean unblocks everything).
blockResourcesarrayOptionalResource types to block.
blockUrlsarrayOptionalURL patterns to block.
cookieActionstringOptionalget, set or clear cookies.
storageActionstringOptionalget, set or clear web storage.
storageTypestringdefault localOptionallocal, session or both.
saveStorageStatePathstringOptionalSave storageState JSON to a path inside the save root.
loadStorageStatePathstringOptionalLoad storageState JSON from a path inside the save root.

Returns

FieldTypeDescription
rules_and_stateobjectActive blocking rules and any cookies or storage that were read.
wait_for_humanBrings the tab to the front for a person to log in or pass a check, waits until a condition holds, then continues in the same session.

Example call

{"name":"wait_for_human","arguments":{"message":"Please sign in and finish the security check, then leave the tab open.","selector":"#account-menu","timeoutMs":120000}}

Parameters

ParameterTypeNeededDescription
messagestringOptionalInstruction shown for the person.
urlMatchesstringOptionalResume when the page URL matches this pattern.
selectorstringOptionalResume when this element appears.
textMatchesstringOptionalResume when the page text matches.
textGonestringOptionalResume when this text disappears.
timeoutMsnumberdefault 120000OptionalHow long to wait; keep it below your client's request timeout.

Returns

FieldTypeDescription
wait_resultobjectWhether the condition was met before the timeout, with the tabId.

What Turbo Browser MCP is for

Turbo Browser MCP is a no-cost MCP server, under the MIT licence, that gives AI agents control of Chromium through Playwright. It groups browser actions into 11 batched tools, returns compact snapshots with only the changed elements, fills a whole form in one call, and blocks Submit, Publish and Send clicks until the agent explicitly allows them.

  • Delta snapshots

    With onlyChanged set, a snapshot lists just the elements added or altered since the last one, so a click costs tokens only for what moved.

  • Whole-form fill in one call

    One request sets text, selects, checkboxes, rich-text editors, shadow DOM and iframe fields, and a Gmail compose box.

  • Stop before the final click

    The last step of a form is refused until the agent opts in, covering Submit, Publish, Send, Apply and payment buttons.

  • Proof and duplicate protection

    From version 0.5.0 an expect check reports submitVerified with evidence, and a submission key keeps a rerun from filing the same entry twice.

Show all 8 features
  • Uploads from disk

    A file goes in by its path, so a PDF never passes through the conversation as base64.

  • Parallel isolated tabs

    tabs_batch opens up to ten URLs at once, each optionally in its own context with separate cookies and storage.

  • Four connection modes

    auto, cdp, persistent and launch let an agent reuse a signed-in Chrome, a saved profile or a clean headless browser.

  • Human hand-off

    wait_for_human pauses for a login or two-factor step and resumes in the same tab when a condition is met.

What people use it for

  • Fill a long web form and stop before the final click

    An agent fills every field and uploads a PDF in one call, then waits for a person to approve the Submit button.

  • Run several logged-in sessions side by side

    Open tabs in separate isolated contexts so each account keeps its own cookies and storage during one automation run.

  • Collect article text with less page weight

    Block images and media with a preset, then pull clean markdown from the main content instead of raw HTML.

  • Triage a broken page

    Read only the failed requests and console errors, then open the headers or body of one request to find the cause.

  • Hand a login or 2FA step to a person

    The agent pauses, the user signs in or passes the check in the visible tab, and the run resumes in the same session.

Strengths and limits

Strengths

  • After a click, the maintainer's benchmark returns about 56 tokens of changed lines against 3,582 for a full accessibility-tree rescan, measured by the author's own script (maintainer README).
  • Form submission is guarded by default and can be verified and de-duplicated; the maintainer comparison table lists no such default guard for Playwright MCP or Chrome DevTools MCP, and Hokai has not tested either.
  • MIT licensed, no account, no API key and no hosted service, so the only cost is the machine and the tokens your agent uses.
  • The tool schema is small: the author measures 3,170 tokens for its tools against 3,917 for Playwright MCP, which matters because the schema is resent on every turn.

Limits

  • Not published on npm: the registry answered package not found when Hokai checked, so the README's npx line does not work yet and you must clone the repository and update it yourself.
  • Very new and run by one maintainer, with two versions in two days. Hokai found no independent write-up or test when it looked, so every benchmark here is the author's own.
  • It does not hide automation, solve CAPTCHAs or use stealth patches, so sites that re-check for bots on every request still see an automated browser and need a human step.
  • It leaves out performance traces, Lighthouse audits and heap snapshots, which the maintainer says Chrome DevTools MCP keeps; teams that need those run both servers.

Turbo Browser MCP alternatives

SkillWorks withConnects byChecked
Turbo Browser MCP This pageClaude · Codex · CursorMCP16 Nov
MantisPick Turbo Browser MCP if you want batched calls, whole-form fill and a default submit guard; pick Mantis if you prefer a local browser that answers in numbered text.Claude · Codex · CursorMCP11 Nov
KernelPick Turbo Browser MCP if you want a free browser on your own machine; pick Kernel if you need hosted cloud Chromium behind a hosted MCP server.Claude · Cursor · Windsurf · VS CodeMCP13 Nov
NotteCloud browsers, browser agents and deployable browser functions for AI, with a CLI and two MCP servers.Claude · Cursor · WindsurfMCP11 Oct

Turbo Browser MCP facts

Protocol
MCP
Latency class
Fast
Open source
Yes
Maintainer founded
2026

What changed

  • New feature v0.5.0: submit evidence, idempotency ledger, modal and network commit guards source
  • Release Release turbo-browser-mcp v0.4.0 source

Questions about Turbo Browser MCP

How much does Turbo Browser MCP cost?

Nothing. The code is MIT licensed, runs on your own machine and needs no account or API key, and the repository lists no paid plan or hosted service (read 16 November 2026). Your only costs are the tokens your agent's model uses and the computer the browser runs on.

Is Turbo Browser MCP really open source, and are there usage limits?

Yes. The repository carries an MIT licence file, and package.json names the same licence. No request quota or rate limit is stated anywhere in the docs, so the practical limits are your machine and the sites you visit.

Will it submit a form on its own?

Not by default. Final actions that look like Submit, Publish, Send, Apply, Confirm or a payment are blocked unless the agent passes allowSubmit or you start the server with --allow-submit. The README admits button labels can mislead, so it adds an optional network guard for POST and PUT requests, switched on with TURBO_BROWSER_NETWORK_GUARD=true and off by default because many multi-step wizards send a request on every Next.

Why does npx turbo-browser-mcp fail for me?

The README shows an npx line, but the npm registry returned package not found when Hokai checked on 16 November 2026, even though package.json and server.json both name the package. Until the maintainer publishes it, clone the repository, run npm install and point your MCP client at src/index.mjs with node. Hokai did exactly that at the newest commit on the same day: the doctor check reported Node, Playwright and a headless Chromium, and a stdio tool-list call returned every tool described on this page.

Can it use the Chrome I am already signed in to?

Yes. In its default auto mode the server first looks for Chrome's remote debugging endpoint at http://127.0.0.1:9222, attaches to that browser, and only then falls back to launching Chromium. A dedicated profile through --launch-chrome avoids the Allow remote debugging prompt that newer Chrome versions show on connection. A site that re-checks for automation on every request will still treat the browser as automated.

About Turbo Browser MCP

Turbo Browser MCP is an open-source Model Context Protocol server by developer Trung Thanh Tran, first published in November 2026, that lets an AI agent drive a real Chromium browser through Playwright. It is a skill, not a tool: you never open it yourself; it is called by Claude Code, Cursor or another MCP client. Its aim is fewer tokens and fewer round trips. Instead of one call per click, it groups form fills, clicks and tab opens into single calls, and after a click it can return only the page elements that changed. The repository showed 106 stars on 16 November 2026.

Read the full background

The server speaks MCP over stdio, needs no account or API key, and runs on Node 20 or newer through playwright-core. It can attach to a Chrome you already have open with remote debugging, reuse a saved profile folder, or start a clean headless browser. Pages come back as compact numbered snapshots, and the server looks inside iframes and open shadow DOM. Files are uploaded from a path on disk rather than pasted into the conversation as base64.

By default the server refuses the last step of a form: buttons that look like Submit, Publish, Send or a payment, and Enter presses inside forms, are blocked until the agent sets allowSubmit, or the server is launched with a startup flag. Version 0.5.0 added a proof check, a duplicate-submission ledger and a modal rule; the maintainer README says readers running agents on real portals reported gaps in label-only checks (read 16 November 2026). The README gives setup for Claude Code, Cursor, Antigravity and Codex; Hokai has not tried other MCP clients.

Hosted browsers with proxies and CAPTCHA solving are what Kernel, Notte and Browserbase sell; this server runs only on your own machine and does not try to hide automation. Mantis is the local rival with the same token-saving aim and answers in numbered text. The maintainer comparison table sets it against Playwright MCP and Chrome DevTools MCP, neither of which has a Hokai page yet; the second keeps Lighthouse audits and heap snapshots that this server leaves out. See the wider browser skills shelf and the best MCP servers by job guide.

Decision: try it on your own machine if you want a local, token-lean browser with a guarded submit and can live with a git clone. Choose a hosted browser if you need proxies, and keep Chrome DevTools MCP beside it if you need Lighthouse.

Sources and guides

More Agent Skills on Hokai

  • OlostepScrape, crawl, map, search and answer from the web, as an API, a CLI and an MCP server.