Stealthy browser automation, testing, and web-scraping via CDP Mode.
WebSkrap is installed from its publisher's own source and answers where it runs, so this marketplace is not in the path of a single call. There is no address here to send one to, and a panel that pretended otherwise would be showing you an answer we made up. Install it and call it from your own client — the Installation tab has the entry for each one.
What it does
Stealth web fetch, search and real Chrome control for AI agents.
Quickstart
# 1 — add it to Claude Code
claude mcp add --scope user webskrap-mcp '--' uvx webskrap mcp
# 2 — ask your agent for something these tools do
# fetch, stealth_fetch, fetch_session_close and 18 more
WebSkrap costs nothing on mcprush: there is no plan to choose here, no cap to set and nothing mcprush can bill you for.
Collected from a public index. Nobody has claimed this account, so nothing here was written by its author — claim it if it is yours.
Where are you running it?
Every route below installs the same thing and ends at the same approval screen. This one runs on your machine: your client starts WebSkrap as a process under your own user, with your files and your network, so the tool surface below is what it can do to your computer rather than to a server somewhere else. Its scan result is on the Security tab — read the surface before you approve it.
This is a public server: you run it yourself and this marketplace is not in the path. You need uv (docs.astral.sh/uv): uvx comes with it. Claude Code registers it in one command; --scope user makes it available in every project. The '--' is in quotes so that PowerShell passes it on.
claude mcp add --scope user webskrap-mcp '--' uvx webskrap mcpReconnect, or start a new session, and the tools appear in the model’s tool list.
One config entry your client uses to start the process locally. A local server runs with your file system and your network, which is why it is priced without metering.
21 tools, with what each one reads, writes and reaches shown before you agree — the same list on every route above. Read the tool surface.
Tool surface
What the model actually sees. Descriptions are diffed on every release — see version history.
Read from its source code. A server that builds its tool list when it starts can list more tools than this.
Fetch a URL in headless Chromium and return page data.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| url | string | yes | The URL to load. |
| profile | string | no | Profile label; the browser keeps its native identity here. |
| channel | string | no | Browser channel, e.g. chrome. If chrome cannot launch (Linux ARM64 has no build), Edge and then bundled Chromium are tried. |
| wait_until | string | no | commit, domcontentloaded, load, or networkidle. |
| resource_policy | string | no | all (default, as in stealth_fetch, so both can share a warm session), auto (lite for text, all for HTML), lite, or documents (also blocks CSS). |
| timeout_ms | number | no | Navigation timeout in milliseconds. |
| max_chars | integer | no | Maximum characters of page text to return. |
| offset | integer | no | Character index to start the returned text at. Pass back the result's next_text_offset to read a long page in windows instead of fetching it again with a bigger max_chars. |
| text_only | boolean | no | Return clean visible text (default) instead of raw HTML. |
| include_links | boolean | no | Also return the page's outbound links, so a crawl does not need a second fetch of the raw HTML. Off by default: on a link-heavy page they cost more than the text. |
| max_links | integer | no | How many links to return; links_total counts them all. |
| decline_cookies | boolean | no | Click a cookie consent notice's reject button after load, so the banner does not bury the page text. |
| ready_selector | string | no | Optional CSS selector or snapshot ref that must be visible before extraction. Uses timeout_ms after consent dismissal. |
| solve_cloudflare | boolean | no | Click Cloudflare's Turnstile checkbox on its "Just a moment" page like a person and wait for the real page, for up to 60 s beyond timeout_ms. Off by default. Use responsibly. |
| session | string | no | Opt-in warm fetch session. Shares cookies/storage across calls; keep profile/config identical and close with fetch_session_close. |
Fetch a URL in Chromium with every stealth control exposed.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| url | string | yes | The URL to load. |
| profile | string | no | Bundled profile; profile_mode decides how much pages see. |
| channel | string | no | Browser channel, e.g. chrome. If chrome cannot launch (Linux ARM64 has no build), Edge and then bundled Chromium are tried. |
| headless | boolean | no | Run headless. Headed is more robust against detection. |
| user_data_dir | string | no | Persistent browser profile directory, relative to the MCP profile root. |
| profile_mode | string | no | native (default; the browser's own identity), locale (locale, timezone, media, headers) or full (also viewport and UA). |
| reduce_fingerprint_surface | boolean | no | Disable WebGL and canvas readback via flags. |
| mask_headless_user_agent | boolean | no | Rewrite the HeadlessChrome UA token to Chrome. Empties high-entropy client hints; prefer virtual_display. |
| virtual_display | boolean | no | Run headed Chromium on a private Xvfb display instead of headless (Linux with Xvfb). No HeadlessChrome token and full client hints, without any override. |
| gpu | no | WebGL renderer. "auto" is Chromium's choice (SwiftShader without a GPU, a known headless tell), and "mesa" under virtual_display without a GPU when lavapipe is installed; "mesa" uses Mesa's lavapipe ( | |
| fake_media_devices | boolean | no | Provide Chromium's synthetic camera and microphone. |
| webrtc_ip_handling_policy | string | no | Chromium WebRTC ICE policy, e.g. disable_non_proxied_udp. |
| timeout_ms | number | no | Navigation timeout in milliseconds. |
| max_chars | integer | no | Maximum characters of page text to return. |
| offset | integer | no | Character index to start the returned text at; pass back next_text_offset to continue. |
| text_only | boolean | no | Return clean visible text (default) instead of raw HTML. |
| include_links | boolean | no | Also return the page's outbound links. |
| max_links | integer | no | How many links to return; links_total counts them all. |
| decline_cookies | boolean | no | Click a cookie consent notice's reject button after load, so the banner does not bury the page text. |
| ready_selector | string | no | Optional CSS selector or snapshot ref that must be visible before extraction. Uses timeout_ms after consent dismissal. |
Release a named fetch session's browser and temporary cookies/storage.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session | string | yes | — |
Search the web and return the organic hits as title, url and snippet.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| query | string | yes | Words to search for. |
| engine | string | no | bing (default) or ddg (DuckDuckGo). |
| max_results | integer | no | How many hits to return; hits_total counts them all. |
| profile | string | no | Bundled profile; profile_mode decides how much pages see. |
| channel | string | no | Browser channel, e.g. chrome. If chrome cannot launch (Linux ARM64 has no build), Edge and then bundled Chromium are tried. |
| headless | boolean | no | Run headless. Headed is more robust against detection. |
| user_data_dir | string | no | Persistent browser profile directory, relative to the MCP profile root. |
| profile_mode | string | no | native (default; the browser's own identity), locale (locale, timezone, media, headers) or full (also viewport and UA). |
| reduce_fingerprint_surface | boolean | no | Disable WebGL and canvas readback via flags. |
| mask_headless_user_agent | boolean | no | Rewrite the HeadlessChrome UA token to Chrome. Empties high-entropy client hints; prefer virtual_display. |
| virtual_display | boolean | no | Run headed Chromium on a private Xvfb display instead of headless (Linux with Xvfb). No HeadlessChrome token and full client hints, without any override. |
| gpu | no | WebGL renderer. "auto" is Chromium's choice (SwiftShader without a GPU, a known headless tell), and "mesa" under virtual_display without a GPU when lavapipe is installed; "mesa" uses Mesa's lavapipe ( | |
| fake_media_devices | boolean | no | Provide Chromium's synthetic camera and microphone. |
| webrtc_ip_handling_policy | string | no | Chromium WebRTC ICE policy, e.g. disable_non_proxied_udp. |
| timeout_ms | number | no | Navigation timeout in milliseconds. |
| decline_cookies | boolean | no | Click a cookie consent notice's reject button on the results page before reading it. |
Report whether WebSkrap can drive a browser here, and how it is configured.
Takes no parameters.
Start (or reuse) a persistent browser session with no visible window.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| url | string | no | Optional URL to open after launch. |
| session | string | no | Session name; letters, digits, '.', '_' or '-'. |
| webrtc_ip_handling_policy | string | no | Chromium WebRTC ICE policy, e.g. disable_non_proxied_udp to keep the host's LAN and direct public addresses out of WebRTC candidates. Fixed at launch; reopening a running session with a different poli |
| fake_media_devices | boolean | no | Supply Chromium's synthetic camera and microphone. |
| decline_cookies | boolean | no | With url, click a consent notice's reject control; cookie_notice_declined reports the strategy or null. |
| virtual_display | boolean | no | Run headed Chromium on a private Xvfb display instead of headless (Linux with Xvfb). No HeadlessChrome token and full client hints. The display lives as long as the session. |
| gpu | no | WebGL renderer. "auto" is Chromium's choice (SwiftShader without a GPU), and "mesa" under virtual_display without a GPU when lavapipe is installed; "mesa" uses Mesa's lavapipe (Linux). The result repo | |
| profile_mode | string | no | native (default; the browser's own identity) or locale (the profile's timezone and languages, set at launch; Linux). |
| profile | string | no | Bundled profile for locale mode; desktop-chrome when null. |
| solve_cloudflare | boolean | no | With url, click Cloudflare's Turnstile checkbox like a person and wait up to 60 s for the real page. Use responsibly. |
List, open, select or close the session's tabs.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| action | string | no | list, new, select, or close. |
| session | string | no | Browser session name. |
| index | integer | no | Tab index from list, counted from 0 in opening order. Needed for select; close without it closes the current tab. |
| url | string | no | For new only: the URL to load in the new tab. |
Navigate the session's current page to a URL.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| url | string | yes | The URL to navigate to. |
| session | string | no | Browser session name. |
| wait_until | string | no | commit, domcontentloaded, load, or networkidle. |
| timeout_ms | number | no | Navigation timeout in milliseconds. |
| decline_cookies | boolean | no | Click a consent notice's reject control after the load; cookie_notice_declined reports the strategy or null. |
| solve_cloudflare | boolean | no | Click Cloudflare's Turnstile checkbox like a person and wait up to 60 s beyond timeout_ms for the real page. Use responsibly. |
Return an aria snapshot of the current page with eN element refs.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session | string | no | Browser session name. |
| depth | integer | no | Maximum snapshot tree depth (default 6; None for unlimited). Lower it before raising max_chars: a shallow tree of the whole page is usually more useful than the first characters of a deep one. |
| max_chars | integer | no | Maximum characters of snapshot text to return. |
| offset | integer | no | Character index to start at; pass back next_snapshot_offset to read the rest of a clipped tree. |
| target | string | no | Snapshot only this ref or unique main-frame CSS selector. |
| boxes | boolean | no | Include viewport-relative CSS pixel boxes. |
Find semantic lines and ancestor context without unrelated subtrees.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| text | string | no | — |
| regex | string | no | — |
| session | string | no | — |
| target | string | no | — |
| max_matches | integer | no | — |
| max_chars | integer | no | — |
| offset | integer | no | — |
Read visible body text without building an accessibility snapshot.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session | string | no | Browser session name. |
| max_chars | integer | no | Maximum characters to return (default 8000). |
| offset | integer | no | Character index to start at; pass back next_text_offset for more. |
| pattern | string | no | Python regex (max 200 chars); (?i) for case-insensitive. max_chars and offset do not apply. |
| context | integer | no | Lines kept before and after each match, 0 to 3. |
| max_matches | integer | no | Matches returned, 1 to 50. |
Interact with an element on the session's current page.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| action | string | yes | click, dblclick, hover, check, uncheck, fill, type, or select. |
| target | string | yes | Snapshot ref (e.g. "e15") or a CSS selector. |
| values | array | no | Value arguments: none for click/dblclick/hover/check/uncheck, exactly one for fill/type, one or more for select. |
| session | string | no | Browser session name. |
| timeout_ms | number | no | Action timeout in milliseconds. |
Press a keyboard key on the session's current page.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| key | string | yes | Key to press, e.g. Enter, Tab, or Control+a. |
| session | string | no | Browser session name. |
| timeout_ms | number | no | Action timeout in milliseconds. |
Wait for one condition on the session's current page.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| text | string | no | Wait until this text is visible on the page. |
| text_gone | string | no | Wait until this text is gone. Text that was never there satisfies the wait immediately. |
| selector | string | no | Wait until this snapshot ref (e.g. "e15") or CSS selector reaches selector_state. |
| selector_state | string | no | attached, detached, visible, or hidden. Used with selector. |
| load_state | string | no | domcontentloaded, load, or networkidle. |
| session | string | no | Browser session name. |
| timeout_ms | number | no | How long to wait before failing with a timeout. |
Screenshot the session's current page to a PNG file.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| path | string | no | Output PNG path relative to the output directory; nested subdirectories are allowed and created. Auto-generated when omitted. |
| session | string | no | Browser session name. |
| full_page | boolean | no | Capture the full scrollable page. |
| target | string | no | Snapshot ref (e.g. "e15", or "f12345e15" inside an iframe) or CSS selector: capture only that element's bounding box, reported as clip. Not combined with full_page. |
Return the rendered viewport as inline PNG image content and metadata.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session | string | no | Browser session name. |
Interact using coordinates from a fresh browser_view image.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| action | string | yes | click, dblclick, move, scroll, or drag. |
| x | number | yes | Horizontal coordinate from the original view's left edge. |
| y | number | yes | Vertical coordinate from the original view's top edge. |
| session | string | no | Browser session name. |
| end_x | number | no | Drag destination x; required with end_y for drag only. |
| end_y | number | no | Drag destination y. |
| delta_x | number | no | Horizontal CSS-pixel wheel delta, scroll only. |
| delta_y | number | no | Vertical CSS-pixel wheel delta; positive scrolls down. |
| button | string | no | left, middle, or right; click, dblclick, and drag only. |
| timeout_ms | number | no | Action timeout in milliseconds. |
Insert literal text after focusing a control with browser_mouse or Tab.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| text | string | yes | Literal text to insert, including Unicode. |
| session | string | no | Browser session name. |
| timeout_ms | number | no | Action timeout in milliseconds. |
Evaluate JavaScript on the session's current page.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| expression | string | yes | JavaScript expression or function to evaluate. |
| session | string | no | Browser session name. |
| timeout_ms | number | no | Action timeout in milliseconds. |
| max_chars | integer | no | Maximum characters of encoded result to return. |
Close a browser session (its profile persists unless delete_data).
| Parameter | Type | Required | Notes |
|---|---|---|---|
| session | string | no | Browser session name; ignored when all_sessions is set. |
| delete_data | boolean | no | Also delete the session's on-disk profile data. |
| all_sessions | boolean | no | Close every session. |
List persistent browser sessions and whether each is running.
Takes no parameters.
- Every tool, no call limit
- No card, and mcprush charges nothing for it
- Source published under Apache-2.0
- Runs on your machine — nothing of it reaches our gateway
- Nothing to cap, because mcprush bills nothing
What counts against your monthly calls
| Tool | Unit | Calls used | Out of the allowance |
|---|
Nothing here is billable: mcprush charges nothing to install WebSkrap or to call it, at any volume.
Two independent axes, because powerful and malicious are different questions. The grade is threat only. The capability level is blast radius, and it is never a penalty on the grade — it is priced as one subtract-only term in the score, where you can see it.
| Term | Level | What it prices | Points |
|---|---|---|---|
| capability-exposure | high | capability blast radius (high) — client exposure if the model is manipulated | −6 |
| verification-discount | source | publisher verification (provenance) — cryptographic build provenance ties the artifact to its source | −0 |
| coverage-honesty | source | inspection depth (source) — how much of the target the scan could see | −0 |
What the scan could actually read
A grade is only as meaningful as its coverage, so the scanner publishes its own depth before it publishes its result.
Tools were statically extracted from the published source (21 recovered), not enumerated from a running server. Tool-poisoning, Unicode-smuggling, capability and toxic-flow analysis ran on this inferred surface, but a mis-parsed registration could be missed or mis-attributed, so tool-derived findings are capped below “confirmed”. To grade the real runtime surface, scan the running server: --command "npx -y <package>".
Capability — what it could do if the model were manipulated
Tags derived from each tool’s schema and the implementation, not from what the tool calls itself. high is the level these add up to.
| Tool | Capability tags | Why the tag was assigned |
|---|---|---|
| fetch | untrusted-input | untrusted-input: keyword "fetch" |
| stealth_fetch | untrusted-input | untrusted-input: keyword "fetch" |
| fetch_session_close | untrusted-input | untrusted-input: keyword "fetch" |
| search | no tags | |
| doctor | untrusted-input | untrusted-input: keyword "fetch" |
| browser_open | no tags | |
| browser_tabs | no tags | |
| browser_goto | no tags | |
| browser_snapshot | no tags | |
| browser_find | no tags | |
| browser_text | no tags | |
| browser_interact | no tags | |
| browser_press | no tags | |
| browser_wait_for | no tags | |
| browser_screenshot | no tags | |
| browser_view | no tags | |
| browser_mouse | no tags | |
| browser_insert_text | no tags | |
| browser_eval | code-exec | code-exec: keyword "eval" in tool name |
| browser_close | no tags | |
| browser_list | no tags |
Toxic-flow graph
The lethal trifecta, checked as a graph rather than as a checklist: untrusted input, a sensitive source and an external sink have to meet before there is a path worth worrying about.
Supply chain and provenance
This is the first scan of this surface here, so there is nothing yet to compare it against.
Every result on this tab comes from one deterministic pass over the published package — offline, rule by rule, and auditable line by line above. Same methodology version, same bytes, same score.
Release history
Always latest — the install command names no version, so it installs the newest release when it runs; the current release on file is 3.6.1. Pin a release below and the command asks for exactly that one.
No release note was published with this version.
- The version this listing was on when mcprush began following its releases. The registry was not asked when it shipped, so this row carries no date.
One review per account, from a signed-in account that does not publish this listing. Nothing else is asked — you can say what you think before you install it. Publishers can reply once per review, and a review can be edited or deleted by whoever wrote it.
Any signed-in account that does not publish WebSkrap can review it, once — before installing it or after. A review from an account that has installed it is marked as one.
Nobody has reviewed this listing. The rating on the card is the mean of the reviews written here and nothing else, so there is no rating until somebody writes the first — which takes a signed-in account that does not publish it, and nothing else.