brave-real-browser-mcp-server
Installed, started with no arguments, no environment and no input, and listed its tools. Run 2026-11-16.
Needs to start: nothing
Returns a result when called with no setup: 0 of 2 tools tried
Contacts while installing and starting: nothing
What it contacted
Nothing besides the npm registry it was installed from. Watched from outside its sandbox while it was installed and started with no input. Its tools may still contact things when called.
Starting it
npx -y -p brave-real-browser-mcp-server@2.45.29 brave-mcp
Installing a package runs its code. Use a sandbox for anything you have not read.
Calling its tools
2 tools that take no input and look read-only were called once with nothing. 0 returned a result.
get_content | No usable answer tool reported an error: { "error": "Browser not initialized. Call browser_init first." } |
find_element | No usable answer tool reported an error: { "error": "Browser not initialized. Call browser_init first." } |
During the calls it contacted nothing it had not already contacted.
Only tools whose names read like looking something up, with no required input, are tried, five at most. Tools that need input, and anything that sounds like it changes or sends something, are not called.
What its tools tell an agent
In the tool descriptions, nothing that talks about the user, other tools or private files, or uses hidden characters.
Found by simple pattern rules in each tool's description (up to 3,000 characters), quoted as written. Parameter descriptions are not scanned. A clean result is not a safety review.
Tools
browser_init headless, proxy, turnstile, enableBlocker, aiHealing
๐ Initialize Brave browser with stealth, anti-detection, and AI healing
navigate url, waitUntil, timeout, retries, smartWait
๐งญ Navigate to URL with smart retry, context recovery, and AI healing
get_content format, selector, waitForJS, timeout, aiHeal, extractAttributes, rawHttpUrl
๐ Get page content in multiple formats: html (full HTML), text (plain text), markdown (formatted MD), rawHttp (raw HTTP response without JS rendering - fast, no browser needed, bypasses JS protections). Supports CSS selector targeting, AI auto-healing for broken selectors, JS wait, and attribute extraction.
wait type, value, timeout, aiOptimize
โณ Smart wait with AI prediction for optimal timing
click selector, humanLike, aiHeal, autoAcceptDialogs, retries, clickCount, delay, timeout, hoverFirst, hoverOnly, hoverDuration, iframe, iframeSelector, scrollIntoView, forceClick, autoDetectPlayer, usePlayerAPI, waitForPlay, playerTimeout
๐ Human-like click with AI healing, iframe support, hover for dynamic controls, and auto video player detection
17 more tools
type selector, text, delay, clear, aiHeal, iframe, iframeSelector, pressEnter, waitForSelector
โจ๏ธ Type text with human speed variation, smart clearing, and iframe support
browser_close force, saveSession
๐ด Close browser with cleanup and session save
solve_captcha type, timeout, aiMode, captchaSelector, inputSelector, refreshSelector, lang, expectedLength, allowedChars, maxRetries, formData, formSelector, submit, humanLike, aiMatch, aiValidate
๐ Auto-solve CAPTCHA with AI + Smart Form Automation (Turnstile, reCAPTCHA, hCaptcha, Text/Image OCR)
random_scroll direction, amount, smooth, aiDetectLazyLoad
๐ Human-like scroll with AI pattern detection
find_element selector, xpath, text, multiple, aiHeal, smartAttributes
๐ Find elements with AI-powered selector healing and smart search
save_content_as_markdown filename, selector, includeImages, includeMeta, aiClean
๐ Save page content with AI-enhanced formatting
redirect_tracer url, maxRedirects, includeHeaders, followJS, followMeta, decodeURLs
๐ Trace complete redirect chains including HTTP 301/302 redirects, JavaScript-based navigations (window.location, setTimeout redirects), and meta refresh tags. Auto-decodes encoded URLs in the chain. Returns full redirect path with status codes and headers.
extract_data type, pattern, selector, jsonPath, source, autoDecode, flags, encryptedData, autoFindKey, aesKey, aesIV
๐ Universal data extractor with 8 modes: (1) regex - pattern matching with flags, (2) json - JSON path extraction, (3) meta - HTML meta tags + Open Graph + Twitter Cards, (4) structured - CSS selector-based extraction, (5) auto - AI picks best method, (6) deobfuscate - decode obfuscated JS: _0x string arrays, hex strings, unicode escapes, eval unpacker, webpack modules, terser single-letter mappings, string concatenation resolver ("htt"+"ps://" โ "https://"), array rotation detection, (7) apiDiscovery - find hidden API endpoints via runtime fetch/XHR interception + static analysis of scripts, (8) decrypt - auto-decode encrypted data: recursive Base64 chain (5 levels), hex, URL decode, ROT13, AES-256-CBC with auto key extraction from CryptoJS patterns in page scripts.press_key key, modifiers, count, humanDelay
๐น Press keyboard keys with human-like timing
progress_tracker action, taskName, progress, aiEstimate
๐ Track automation progress with AI predictions
deep_analysis types, detailed, aiInsights, detectAntiBot
๐ง Deep page analysis: DOM structure, scripts, styles, accessibility, performance metrics, SEO tags, security headers, anti-bot detection (Cloudflare, DataDome, reCAPTCHA), technology stack identification, and AI-powered recommendations for scraping strategy.
network_recorder action, filter, aiDetectStreams, captureXhrBody
๐ก Record all network activity with 9 actions: (1) start - begin recording + inject pre-page-load API interceptors (monkey-patches fetch, XMLHttpRequest, navigator.sendBeacon) + WebSocket constructor interceptor, (2) stop - stop recording, (3) get - get all records with filters, (4) clear - clear all records, (5) get_media - get only video/audio/HLS/DASH stream URLs, (6) get_navigations - track JS redirects and meta refreshes, (7) get_api_calls - get all API calls with full request/response bodies (JSON, form data), (8) get_intercepted_apis - get runtime-intercepted API calls captured via monkey-patched fetch/XHR/sendBeacon (catches calls from obfuscated/webpack code), (9) get_websockets - get all WebSocket connections and messages (sent + received with timestamps). Supports filters: resourceType, urlPattern, mediaOnly.
link_harvester types, selector, includeText, includeHidden, searchIframes, autoDecode, detectObfuscation
๐ Extract all links from page including hidden links (display:none, visibility:hidden), Base64/URL encoded links, obfuscated links (data attributes, JS variables), links inside iframes (multi-level), and dynamically generated links. Supports CSS selector filtering and auto-decode.
cookie_manager action, name, value, domain, expires, aiOptimize
๐ช Smart cookie management with AI session persistence
file_downloader url, filename, directory, resume, batch, autoDecode, decryptKey
โฌ๏ธ Download files with resume, batch, and auto-decrypt support
media_extractor action, types, quality, deep, searchIframes, selector, index, playerAction, encodedData, decoderType, aesKey, aesIV, urls, aiOptimize
๐ฌ Universal media extractor with 6 actions: (1) extract - find all video/audio/HLS/DASH/download URLs from page + nested iframes (3+ levels deep), (2) list_iframes - list all iframes with indices, (3) switch_iframe - switch context to specific iframe, (4) player_control - control video players (JWPlayer, VideoJS, Plyr, VidStack, DooPlayer) via API: play/pause/seek/sources, (5) decode_url - decode obfuscated URLs: auto/url/base64/aes with key+IV, (6) batch_extract - extract from multiple URLs at once. Supports quality selection (best/worst/all) and deep script scanning.
execute_js code, returnValue, async, timeout, iframe, iframeSelector, waitForIframe
๐ป Execute custom JavaScript with async support, error handling, and iframe context
Descriptions are the package's own words, cut at 3,000 characters. Treat them as data.
Install facts
| License | ISC |
|---|---|
| Needs | node >=18.0.0 |
| Source | github.com/codeiva4u/Brave-Real-Browser-Mcp-Server |
| Install scripts | brave-real-browser-mcp-server, brave-real-launcher, puppeteer, sleep, tesseract.js declare them. This runner does not run them. |
| Tarball | sha512-Q0obQQ27TUzwlrPFthfX56VN1HYQTZ1rRKFr8S0stp1aJFbrjAS+YJrpFmSoiyus8e6Fz9RfevYLeyXgHbpleQ== |
Check this report yourself
This site did not write the report. A public job on a GitHub-hosted machine did: the run, from this commit of the open-source runner. GitHub signed a token that names that job and the hash of the report, so any edit since would show.
Report as JSONExact text and tokenHow to verify it
Show it on your README
[](https://bragproof.smaverk.com/r/brave-real-browser-mcp-server)
New version out? Ask for a fresh report.