MCP server that scans web pages for WCAG accessibility issues with axe-core, so AI agents can audit and fix accessibility. By
Free
Nothing to run from here
ThinkRun 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.
Open Installation—
p95 latency
—
call success
local
runs on your machine
—
calls last 7d
What it does
Connects AI coding agents to the user's real Chrome browser with real sessions and cookies so they can test their own frontend work.
Quickstart
# 1 — run it from where its publisher ships it
npx -y @thinkrun/mcp
# 2 — ask your agent something
> Connects AI coding agents to the user's real Chrome browser with real sessions and cookies so they can test their own frontend work.
ThinkRun is free: there is no plan to choose, no cap to set and nothing that can bill you.
Publishernot claimed
THAll 1 server
@thinkrun1 server
1 category
Collected from a public index. Nobody has claimed this account, so nothing here was written by its author — claim it if it is yours.
Installation
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 ThinkRun 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. It is scanned, signed and pinned to the version you choose — read the surface before you approve it.
Claude Codedeclared by this listing
This is a public server: you run it yourself and this marketplace is not in the path. Claude Code registers it in one command.
claude mcp add thinkrun-mcp -- npx -y @thinkrun/mcpReconnect, or start a new session, and the tools appear in the model’s tool list.
Free to install, and it runs from its publisher’s own package: no account, no key, no card.
What it writes
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.
What you approve
66 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.
Create a new browser session. Returns a sessionId to use with all other tools.
Takes no parameters.
Example callreturns →
Get the current status of a browser session including URL, title, and action count.
Takes no parameters.
Example callreturns →
Terminate a browser session and clean up resources. Artifacts are preserved.
Takes no parameters.
Example callreturns →
List all active browser sessions for this user.
Takes no parameters.
Example callreturns →
List artifacts (screenshots, recording) for a session.
Takes no parameters.
Example callreturns →
Poll session status until the session is ready (running/active) or timeout.
Takes no parameters.
Example callreturns →
Set the default session for subsequent tool calls so you do not need to pass sessionId every time.
Takes no parameters.
Example callreturns →
Enable or disable local audit mode for the current local tab-backed session.
Takes no parameters.
Example callreturns →
Switch the MCP server between local (Chrome extension) and cloud (ThinkRun API) modes at runtime.
Takes no parameters.
Example callreturns →
Return the authenticated hosted remote-agent identity and current user-link status.
Takes no parameters.
Example callreturns →
Request a user-approved hosted remote browser-control grant for the current remote agent.
Takes no parameters.
Example callreturns →
Poll the status of a hosted remote browser-control request and any approved grant.
Takes no parameters.
Example callreturns →
Cancel a pending hosted remote browser-control request or revoke an active grant.
Takes no parameters.
Example callreturns →
Stateless: fetch rendered HTML for a URL without a browser session (no sessionId).
Takes no parameters.
Example callreturns →
Stateless: extract plain text for a URL without a browser session. POST /api/cache/text.
Takes no parameters.
Example callreturns →
Stateless: capture a PNG for a URL without a session. Returns JSON with artifactUrl when Accept is application/json.
Takes no parameters.
Example callreturns →
Navigate the browser to a URL. Waits for page load before returning.
Takes no parameters.
Example callreturns →
Navigate back to the previous page in browser history.
Takes no parameters.
Example callreturns →
Navigate forward to the next page in browser history.
Takes no parameters.
Example callreturns →
Wait for an element to appear on the page. Useful after navigation or clicking
Takes no parameters.
Example callreturns →
Click an element on the page using a CSS selector.
Takes no parameters.
Example callreturns →
Click at viewport-relative CSS pixel coordinates. Local mode only.
Takes no parameters.
Example callreturns →
Type text into an input, textarea, or contenteditable element; local extension mode also supports role=textbox.
Takes no parameters.
Example callreturns →
Direct replacement for an input, textarea, or contenteditable element; local extension mode also supports role=textbox.
Takes no parameters.
Example callreturns →
Press a keyboard key. Common keys: Enter, Tab, Escape, ArrowDown, ArrowUp, Backspace.
Takes no parameters.
Example callreturns →
Scroll the page or a specific element. Positive y scrolls down, negative scrolls up.
Takes no parameters.
Example callreturns →
Hover the mouse over an element. Useful for triggering tooltips or dropdown menus.
Takes no parameters.
Example callreturns →
Select an option from a <select> dropdown element.
Takes no parameters.
Example callreturns →
Get an accessibility tree snapshot of the current page. This is a lightweight
Takes no parameters.
Example callreturns →
Capture a screenshot of the current page. Returns an image plus any available artifact metadata.
Takes no parameters.
Example callreturns →
Extract content from the page. Types:
Takes no parameters.
Example callreturns →
Execute JavaScript in the browser page context. The script runs in the page,
Takes no parameters.
Example callreturns →
List all open browser tabs. Local mode only.
Takes no parameters.
Example callreturns →
Open a new browser tab, optionally navigating to a URL. Local mode only.
Takes no parameters.
Example callreturns →
Close a browser tab by ID. If no tabId is provided, closes the current tab. Local mode only.
Takes no parameters.
Example callreturns →
Attach to an existing local Chrome tab by tab ID and make it the default session. Local mode only.
Takes no parameters.
Example callreturns →
Switch the active local Chrome tab without rebinding the current default session. Local mode only.
Takes no parameters.
Example callreturns →
Bring the currently bound local browser tab/window to the foreground without rebinding the default session. Local mode only.
Takes no parameters.
Example callreturns →
Open a new local Chrome window and bind the resulting tab as the default session. Local mode only.
Takes no parameters.
Example callreturns →
Release the current local tab ownership without closing the tab. Clears the default session when releasing the bound local session. Local mode only.
Takes no parameters.
Example callreturns →
Return structured local continuity, ownership, reclaim, and bridge recovery diagnostics. Local mode only.
Takes no parameters.
Example callreturns →
Trigger the bounded native-host reset-connection recovery path used for false local circuit-breaker trips. Local mode only.
Takes no parameters.
Example callreturns →
Return the current page URL for the active session.
Takes no parameters.
Example callreturns →
Return the current page title for the active session.
Takes no parameters.
Example callreturns →
Return the current page HTML for the active session. Large documents are truncated explicitly in the response.
Takes no parameters.
Example callreturns →
Pause for a bounded duration in milliseconds (max ${MAX_SLEEP_MS}).
Takes no parameters.
Example callreturns →
Check for pending browser dialogs (alert, confirm, prompt) and view dialog history.
Takes no parameters.
Example callreturns →
Accept or dismiss a pending browser dialog (alert, confirm, prompt).
Takes no parameters.
Example callreturns →
Wait for specific text to appear on the page.
Takes no parameters.
Example callreturns →
Get browser console messages (log, warn, error) from the current session.
Takes no parameters.
Example callreturns →
Get network requests made by the browser during this session.
Takes no parameters.
Example callreturns →
Clear console and network logs for the session.
Takes no parameters.
Example callreturns →
Run a direct native-host action against the currently attached local tab.
Takes no parameters.
Example callreturns →
Check the status of a previously started local action. Local mode only.
Takes no parameters.
Example callreturns →
Cancel a running local action. Local mode only.
Takes no parameters.
Example callreturns →
Create an AI task plan from natural language. The plan breaks your request into
Takes no parameters.
Example callreturns →
Execute a previously created task plan. Cloud mode only. Unavailable in remote-agent-only mode (requires THINKRUN_API_KEY).
Takes no parameters.
Example callreturns →
Check the status of a task plan execution. Cloud mode only. Unavailable in remote-agent-only mode (requires THINKRUN_API_KEY).
Takes no parameters.
Example callreturns →
Accessibility/DOM snapshot of the approved session tab.
Takes no parameters.
Example callreturns →
Extract text/HTML/attributes from the approved tab.
Takes no parameters.
Example callreturns →
Click an element in the approved tab.
Takes no parameters.
Example callreturns →
Fill a form field (non-sensitive) in the approved tab.
Takes no parameters.
Example callreturns →
Append text to an editable element in the approved tab, or replace its content with clear=true, with an optional per-character delay.
Takes no parameters.
Example callreturns →
Wait for text to appear in the approved tab.
Takes no parameters.
Example callreturns →
Capture a screenshot of the approved tab.
Takes no parameters.
Example callreturns →
Navigate the approved tab to a URL (subject to sensitive-domain policy).
Takes no parameters.
Example callreturns →
Freeyour plan
$0 forever
- Every tool, no call limit
- No card, no account needed
- Source published under a licence you can read
- Runs on your machine — nothing of it reaches our gateway
- Nothing to cap, because nothing bills
What counts against your monthly calls
| Tool | Unit | Calls used | Out of the allowance |
|---|
Nothing here is billable. ThinkRun costs nothing to install and nothing to call, at any volume.
Call allowance · this installenforced at the gateway
Monthly allowance, in calls
—
—
Alert at 80% · then the tool returns a normal MCP error
Used this month: $8.42
$16.58 left
ATrust score 93 / 100
Release 0.3.13 · scanned 25 Aug 2026
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.
Threat score100
Client score93
Findings4 threat · 5 capability
Capabilityhigh
Coveragesource
Verificationrepo
Surface digest93cac63c459f3db1bf387921620d0246354134a3b0c4460c7155c944b0a217f2
Score vectorthreat 100 → client 93
| Term | Level | What it prices | Points |
|---|---|---|---|
| capability-exposure | high | capability blast radius (high) — client exposure if the model is manipulated | −6 |
| verification-discount | repo | publisher verification (public source) — no provenance, but the source is public and inspectable | −1 |
| coverage-honesty | source | inspection depth (source) — how much of the target the scan could see | −0 |
score = clamp(round(threatScore − E_cap − E_ver − E_cov), 0, 100). No model, no opinion, no network call at scoring time: the same methodology version against the same bytes always produces this number, and every line above is a field of the report rather than a summary of it.
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.
Tool surfaceread
Implementation sourceread
Package metadataread
Live transportnot available
Tools were statically extracted from the published source (66 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 |
|---|---|---|
| agent_identity | no tags | |
| browser_request_control | no tags | |
| browser_grant_status | no tags | |
| browser_release_control | no tags | |
| browser_snapshot | no tags | |
| browser_extract | no tags | |
| browser_click | no tags | |
| browser_fill | no tags | |
| browser_type_text | no tags | |
| browser_wait_for_text | no tags | |
| browser_screenshot | no tags | |
| browser_navigate | no tags | |
| session_create | no tags | |
| session_status | no tags | |
| session_delete | no tags | |
| session_list | no tags | |
| session_artifacts | no tags | |
| session_wait_ready | no tags | |
| session_use | no tags | |
| set_auditing | no tags | |
| set_mode | no tags | |
| page_cache_html | untrusted-input | |
| page_cache_text | no tags | |
| page_cache_screenshot | no tags | |
| navigate | no tags | |
| go_back | no tags | |
| go_forward | no tags | |
| wait_for_element | no tags | |
| click | no tags | |
| click_at | no tags | |
| type_text | no tags | |
| fill | no tags | |
| press_key | no tags | |
| scroll | no tags | |
| hover | no tags | |
| select_option | no tags | |
| snapshot | no tags | |
| screenshot | no tags | |
| extract | code-exec | |
| evaluate | code-exec | |
| tab_list | no tags | |
| tab_new | no tags | |
| tab_close | no tags | |
| tab_attach | no tags | |
| tab_switch | no tags | |
| focus | no tags | |
| window_new | no tags | |
| session_release | no tags | |
| local_diagnostics | no tags | |
| local_reset_connection | no tags | |
| get_url | no tags | |
| get_title | no tags | |
| get_html | no tags | |
| sleep | no tags | |
| get_dialog | no tags | |
| handle_dialog | no tags | |
| wait_for_text | no tags | |
| console_messages | no tags | |
| network_requests | no tags | |
| clear_logs | no tags | |
| local_action_run | no tags | |
| local_action_status | no tags | |
| local_action_cancel | no tags | |
| task_create | no tags | |
| task_execute | no tags | |
| task_status | 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.
Untrusted input can drive an external action even though no sensitive source is exposed.
Supply chain and provenance
Integrityfirst-seen
Surface digest93cac63c459f3db1
Scanned2026-08-25
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
Pinned to 0.3.13 — the install command below asks for that release. A pin is part of an install, so it is kept for this visit and written down when you install.
0.3.13
undated
installedNo release note was published with this version.
—
no reviews yet
Reviews are earned
Only accounts with at least 50 real tool calls against this server in the last 90 days can post. Ratings are weighted by how much the reviewer actually uses it, and publishers can reply once per review.
Reviews are earned
Writing one takes an account with at least 50 real tool calls against ThinkRun in the last 90 days. That is the whole gate — there is no other way to post, which is why the counts beside each review are worth reading.
No reviews yet
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 an account with 50 real tool calls against it.