Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@browser-research-mcp visit ppac.gov.in and get the latest crude oil price"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Browser-based research as an MCP server. Drives a real Chromium via
patched Playwright (patchright) so the agent can read JavaScript-rendered
tables, dynamic charts, login-walled dashboards, and AJAX dropdowns that
the cheaper rungs of the fetch ladder can't reach.
This is the last rung of the ladder:
web_search → web_fetch → pdf_fetch → http_post_form → browser-researchTool | Purpose |
| Open a URL with Chromium, return DOM text + screenshot. Cheap, no LLM call. |
|
|
Related MCP server: Browser MCP
patchright is a drop-in
Apache-2.0 patched Playwright that disables the AutomationControlled blink
feature, removes Runtime.enable leaks, and a few other detection vectors.
Indian government dashboards (PPAC, RBI, MoSPI, SEBI) work fine with this
without paying for residential proxies or a hosted browser SaaS.
uv tool install browser-research-mcp --python 3.12
# Install Chromium for patchright (one-off):
uv tool run patchright install chromium
# stdio (Claude Desktop / Cursor / desktop clients):
ANTHROPIC_API_KEY=… uvx browser-research
# HTTP (the platform backend):
ANTHROPIC_API_KEY=… uvx browser-research --transport streamable-http --port 7862Var | Required | Default |
| for | — |
| no |
|
| no |
|
| no |
|
| no |
|
Layer | Library |
Browser engine |
|
Structured extraction | Anthropic Claude Sonnet 4.6, with vision input |
MCP transport |
|
Session isolation | One Playwright context per MCP |
Drive a real Chromium through a sequence of steps, then run Sonnet structured extraction on the final state.
Use this when the data is BEHIND an interaction — a Year/Month dropdown
that fires AJAX inline, a tab to click, a "Load more" button, a form
to submit. visit and extract only read the page as it loaded;
act clicks/types/selects first.
Steps are a list of single-key dicts: {"click": "css-selector"} {"fill": {"selector": "#q", "value": "x"}} {"select": {"selector": "#year", "value": "2024-2025"}} {"press": {"selector": "#q", "key": "Enter"}} {"scroll": {"to": "bottom"|"top"|}} {"wait_for_selector": "css-selector"} {"wait_for_load_state": "networkidle"|"load"} {"wait_ms": 1500} {"goto": "https://…"} // mid-flow navigation {"screenshot": {"name": "after-select"}} // logged, not returned
Example — pull PPAC FY2024-25 monthly consumption (a flow that needs the year dropdown change to fire an AJAX request): act( url="https://ppac.gov.in/consumption/products-wise", steps=[ {"wait_for_selector": "#financialYear"}, {"select": {"selector": "#financialYear", "value": "2024-2025"}}, {"wait_for_load_state": "networkidle"}, {"wait_ms": 2000}, ], focus="FY2024-25 monthly LPG, MS, HSD, ATF consumption", )
Returns the same shape as extract PLUS step_results (per-step
timing + ok/error) and final_url.
Args: url: Starting page URL. steps: Ordered list of action dicts (vocabulary above). focus: Extraction focus passed to Sonnet. timeout_ms: Per-step navigation / wait timeout. full_page_screenshot: Whether the final screenshot is full-page.
Returns: {url, domain, title, dateline, summary, key_facts[], numeric_values[], dates[], tables_summary[], step_results[], final_url, kind: "browser"}.
| Name | Required | Description | Default |
|---|---|---|---|
| url | Yes | ||
| steps | Yes | ||
| focus | No | ||
| timeout_ms | No | ||
| full_page_screenshot | No |
| Name | Required | Description |
|---|---|---|
No output parameters | ||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description carries full burden. It discloses that the tool drives a real browser, performs steps, takes screenshots (logged, not returned), and returns step_results and final_url. It clearly states the behavioral traits beyond the schema, such as per-step navigation and extraction focus.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is long but well-structured. It opens with the core purpose, followed by when to use, step vocabulary, example, argument list, and return shape. While every section adds value, some verbosity could be trimmed (e.g., the example could be condensed). However, it is front-loaded with key information and highly readable.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the tool's complexity (5 parameters, no annotations, but output schema exists), the description is exceptionally complete. It explains the return shape (same as extract plus step_results and final_url), covers all parameters, and provides detailed step semantics. No gaps remain for an AI agent to correctly invoke this tool.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 0%, so the description must compensate. It does so thoroughly: describing 'url' as starting page URL, 'steps' as ordered list of action dicts with full vocabulary and examples, 'focus' as extraction focus passed to Sonnet, 'timeout_ms' as per-step timeout, and 'full_page_screenshot' as whether final screenshot is full-page. The example illustrates parameter usage effectively.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states 'Drive a real Chromium through a sequence of steps, then run Sonnet structured extraction on the final state.' It uses a specific verb ('drive', 'run extraction') and resource ('Chromium', 'pages'). It distinguishes from siblings by stating that 'visit' and 'extract' only read the page as loaded, whereas 'act' interacts first.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description explicitly tells when to use this tool: 'Use this when the data is BEHIND an interaction—a Year/Month dropdown that fires AJAX inline, a tab to click, a "Load more" button, a form to submit.' It contrasts with siblings: 'visit and extract only read the page as it loaded; act clicks/types/selects first.' It also provides an extensive step vocabulary and a concrete example.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Visit a URL → focused Sonnet structured extraction.
Sends BOTH rendered text AND a screenshot to Sonnet — so numbers drawn via canvas / SVG (chart values on PPAC, RBI, NSE dashboards) that don't appear in the DOM still get extracted. Same returned shape as pdf_fetch_structured / web_fetch_structured on authority-web-search-mcp.
Args: url: The page URL. focus: What to extract, e.g. "monthly LPG, MS, HSD consumption for FY2024-25" or "Q4 FY26 EBITDA margin and revenue". wait_for_selector: Optional CSS selector to await (see visit). full_page_screenshot: Default True so charts below the fold are seen.
Returns: {url, domain, title, dateline, summary, key_facts[], numeric_values[], dates[], tables_summary[], kind: "browser"}.
| Name | Required | Description | Default |
|---|---|---|---|
| url | Yes | ||
| focus | No | ||
| wait_for_selector | No | ||
| full_page_screenshot | No |
| Name | Required | Description |
|---|---|---|
No output parameters | ||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description carries the full burden. It discloses that both rendered text and screenshot are sent, enabling extraction of non-DOM numbers. However, it omits potential side effects like latency, cost, or limitations.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is well-structured and front-loaded, but the return type listing could be slightly more concise. Still efficient and easy to parse.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the tool's complexity (4 params, output schema), the description covers purpose, parameters, and return shape. However, it does not address error handling or comparison with siblings, leaving some gaps.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Input schema has 0% description coverage, but the description's 'Args' section fully explains each parameter with examples and defaults, adding substantial value beyond the schema.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
Description clearly states 'Visit a URL → focused Sonnet structured extraction', specifying the action and resource. It distinguishes by mentioning extraction of canvas/SVG content, but does not explicitly differentiate from sibling tools 'act' and 'visit'.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description provides examples of what to extract (e.g., 'monthly LPG, MS, HSD consumption') and implies usage for focused extraction, but lacks explicit when-to-use or when-not-to-use compared to siblings.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Open a URL with a real Chromium and return its rendered state.
Use when the cheaper fetch tools (web_fetch, pdf_fetch, http_post_form) fail because the page is a SPA, JS-rendered chart, login-walled, or has a dropdown that's not a separate URL.
Args: url: The page URL. wait_for_selector: Optional CSS selector to await before reading the DOM. Use when data appears only after an AJAX call returns — e.g. ".chart svg", "table#monthly tbody tr". wait_extra_ms: Extra settle time after the wait fires (default 1500). timeout_ms: Hard navigation timeout (default 45s). screenshot: Whether to capture a PNG (default True). Adds ~200ms. full_page_screenshot: Scroll-stitch the whole page (default False). text_cap: Cap on extracted text length (default 30000).
Returns: {url, title, domain, text, screenshot_b64, screenshot_bytes, fetched_at, current_date}
| Name | Required | Description | Default |
|---|---|---|---|
| url | Yes | ||
| wait_for_selector | No | ||
| wait_extra_ms | No | ||
| timeout_ms | No | ||
| screenshot | No | ||
| full_page_screenshot | No | ||
| text_cap | No |
| Name | Required | Description |
|---|---|---|
No output parameters | ||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description carries full burden. It discloses Chromium usage, rendering behavior, performance notes (e.g., screenshot adds ~200ms), and return structure. Minor gap: no mention of idempotency or side effects, but it's clear this is a read-only operation.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Well-structured with a concise purpose statement, bullet-style parameter list, and return documentation. Slightly verbose in parameter descriptions but overall efficient and front-loaded.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Covers all parameters, explains returns, and provides context for use relative to siblings. No mention of rate limits or concurrency, but given the output schema and detailed parameter docs, it is sufficiently complete for effective tool use.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Despite 0% schema coverage, the description provides detailed explanations for all 7 parameters, including semantics, defaults, and usage examples (e.g., 'wait_for_selector: Optional CSS selector to await before reading the DOM'). Fully compensates for missing schema descriptions.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states 'Open a URL with a real Chromium and return its rendered state' and distinguishes from cheaper fetch tools, citing specific use cases like SPAs, JS-rendered charts, and login-walled pages.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Explicitly says 'Use when the cheaper fetch tools fail' and lists alternative tools (web_fetch, pdf_fetch, http_post_form) with concrete scenarios, providing excellent when-to-use guidance.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections. Dates show when Glama detected each change.
v0.1.0actextractvisitEach tool has a clearly distinct purpose: 'visit' fetches raw page state, 'extract' performs structured extraction on a static page, and 'act' handles interactive flows with extraction. The descriptions explicitly highlight when to use each, leaving no ambiguity.
All tool names are single imperative verbs ('act', 'extract', 'visit'), following a consistent pattern. While not verb_noun compound names, the style is uniform and predictable.
With only 3 tools, the server covers the essential browser automation tasks: raw page access, static extraction, and interactive extraction. The count is well-scoped and avoids unnecessary bloat.
The tool set covers the core workflows (fetch, extract, interact+extract). A minor gap is the lack of a standalone interaction tool without extraction, but this can be approximated via 'act' with a trivial focus. Overall, the surface feels reasonably complete for browser research.
Unclaimed servers have limited discoverability.
If you are the server author, to access and configure the admin panel.
Stealth web browser for agents: search, fetch, click, download and type in persistent MCP sessions.
Browser MCP for logged-in tasks. Uses your Chrome — credentials stay local. Zero-token replay.
Scrape, crawl and search the web for AI agents via MCP.
One MCP for the Web. Easily search, crawl, navigate, and extract websites without getting blocked.…
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/rsi-ai-platform/browser-research-mcp'
If you have feedback or need assistance with the MCP directory API, please join our Discord server
