Stealthy browser automation, testing, and web-scraping via CDP Mode.
Browsa 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
Cloud browsers for AI agents: run any web task in plain English on Phantom, Browsa's own browser.
Quickstart
# 1 — add it to Claude Code
claude mcp add --scope user --env 'BROWSA_API_KEY=${BROWSA_API_KEY}' --transport stdio browsa-mcp '--' npx -y @browsa/mcp
# 2 — ask your agent for something these tools do
# run_task, get_job, cancel_job and 17 more
It needs BROWSA_API_KEY: this reads it from your environment, so export it first.
Browsa costs nothing on mcprush: there is no plan to choose here, no cap to set and nothing mcprush can bill you for.
Collected from a public index. Nobody has claimed this account, so nothing here was written by its author — claim it if it is yours.
Where are you running it?
Every route below installs the same thing and ends at the same approval screen. This one runs on your machine: your client starts Browsa as a process under your own user, with your files and your network, so the tool surface below is what it can do to your computer rather than to a server somewhere else. Its scan result is on the Security tab — read the surface before you approve it.
This is a public server: you run it yourself and this marketplace is not in the path. You need Node.js — the current LTS (24) is the safe choice; npx comes with it. Claude Code registers it in one command; --scope user makes it available in every project. Claude Code reads BROWSA_API_KEY (Browsa API key from https://app.browsa.io/dashboard/keys) from your environment each time it starts the server, so the value stays out of its config: export it in your shell profile (macOS, Linux), or run setx BROWSA_API_KEY … and open a new terminal (Windows). The '--' is in quotes so that PowerShell passes it on.
claude mcp add --scope user --env 'BROWSA_API_KEY=${BROWSA_API_KEY}' --transport stdio browsa-mcp '--' npx -y @browsa/mcpReconnect, or start a new session, and the tools appear in the model’s tool list.
It reaches a system of yours, so it needs your own credential rather than the publisher’s: BROWSA_API_KEY — Browsa API key from https://app.browsa.io/dashboard/keys. You set it in your own client’s config, on your machine — this marketplace never holds it.
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.
20 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.
Run a browser-driven AI task on a fresh Phantom cloud browser.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| task | string | yes | Plain-English instructions for the browser agent. Be specific about what to extract and the desired output shape (JSON, list, etc.). Example: 'Go to news.ycombinator.com and return the top 3 story tit |
| llm | string | no | — |
| reasoning | no | — | |
| llm_key_id | string | no | Id of a saved LLM key of this workspace (see list_llm_keys). Default: …. Usually not needed: with no key given, the workspace's default saved key for the provider of llm is used. |
| llm_api_key | string | no | Per-task LLM provider key. Rarely needed: prefer a saved key (llm_key_id, or the workspace default) so the secret never passes through the chat. Overrides llm_key_id. |
| country | string | no | 2-letter egress country (e.g. US, GB, DE). Default: …. |
| max_steps | integer | no | Max agent steps. Default 25. Increase for complex multi-page tasks. |
| initial_url | string | no | Optional pre-navigation URL (saves the first agent step). |
| ja3_profile | string | no | TLS handshake spoof profile (e.g. macos_chrome_137). Useful for TLS-fingerprinted sites. |
| profile_id | string | no | Id of a persistent identity (agt_..., from list_identities). If set, reuses that browser identity (cookies, history, fingerprint stay sticky). Omit for a fresh burner. |
| timeout_minutes | integer | no | Max wall-clock minutes to wait before returning. Default 10. |
Fetch the current state of a previously-submitted job by UUID. Returns
| Parameter | Type | Required | Notes |
|---|---|---|---|
| job_id | string | yes | Job UUID returned by run_task or list_jobs. |
| wait_seconds | integer | no | If set (1-120), block server-side until the job reaches a terminal state or N seconds elapse. |
Cancel a queued or running job. The job ends as failed with error_code CANCELLED and keeps its
| Parameter | Type | Required | Notes |
|---|---|---|---|
| job_id | string | yes | Job UUID. |
Answer an input_required prompt (CAPTCHA solution, human decision point). Flips the job back to running.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| job_id | string | yes | — |
| response | string | yes | Your answer to the agent's question. |
Start a real cloud browser and return its CDP URL, so you can drive it yourself with Playwright or
| Parameter | Type | Required | Notes |
|---|---|---|---|
| ttl | string | no | How long to keep it, e.g. "5m" or "1h". Default 10m. |
| country | string | no | ISO-2 country, e.g. "us". Takes effect only with a proxy. |
| os_type | string | no | Operating system the browser presents, e.g. "macos", "windows". |
| profile_id | string | no | Reuse a saved identity (see list_identities) so the browser starts with its cookies and logins. |
| proxy | object | no | One of two forms. Your own proxy: give host and port, or server as "host:port" / "socks5://host:port", plus username and password when it needs them. Or a tier of Browsa's built-in proxy network, bill |
Say whether Browsa's built-in proxy network is enabled on this server and, when it is, list its
No parameter schema on file.
Stop a browser started with start_browser and release it. Always call this when finished; an unused
| Parameter | Type | Required | Notes |
|---|---|---|---|
| browser_id | string | yes | The id returned by start_browser. |
Read the page of a browser started with start_browser, as compact text: one line per element,
| Parameter | Type | Required | Notes |
|---|---|---|---|
| browser_id | string | yes | The id returned by start_browser. |
| scope | string | no | viewport (default): what is in the viewport, the rest counted. page: the whole page. |
| budget | integer | no | Approximate tokens for the view. Default 2000. |
| expand | integer | no | Id of a container, table, list or select to list in full. |
| tab | string | no | Tab id, from the tabs line of an earlier observe. Default: the tab used last. |
Perform one action in a browser started with start_browser and get back what changed (~ changed,
| Parameter | Type | Required | Notes |
|---|---|---|---|
| browser_id | string | yes | The id returned by start_browser. |
| action | string | yes | — |
| id | integer | no | Target element id, from observe. |
| role | string | no | With name: the first element with this role and name, instead of id. |
| name | string | no | — |
| x | number | no | click: a viewport point instead of an element (with y). |
| y | number | no | — |
| text | string | no | type: the text. wait for text: the text. dialog: the prompt answer. |
| clear | boolean | no | type: empty the field first. |
| submit | boolean | no | type: press Enter afterwards. |
| keys | string | no | press: Enter, Escape, Control+A, or a sequence such as "Tab Tab Enter". |
| option | string | no | select: option text or value. |
| double | boolean | no | click: double click. |
| right | boolean | no | click: right button. |
| dy | integer | no | scroll: pixels, positive is down. |
| to | string | no | scroll. |
| url | string | no | navigate. |
| ms | integer | no | wait: a fixed time in milliseconds, at most 30000. |
| for | string | no | wait: for text to appear, an element to go, or the URL to change. |
| timeout | integer | no | wait: milliseconds, default 10000, at most 30000. |
List this workspace's browsers (burners and saved identities) with their status.
No parameter schema on file.
Get the current credit balance + available top-up packs.
No parameter schema on file.
List webhook subscriptions for this workspace.
No parameter schema on file.
Register a callback URL the platform will POST lifecycle events to (HMAC-SHA256 signed via X-Agents-Signature).
| Parameter | Type | Required | Notes |
|---|---|---|---|
| url | string | yes | HTTPS endpoint you control. |
| events | array | no | Event filter. Omit/empty for ALL events. Examples: job.completed, job.failed, agent.started. |
Delete (disable) a webhook subscription by id.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| webhook_id | string | yes | — |
List persistent browser identities (long-lived cookies/history/fingerprint, billed monthly). Use one in
No parameter schema on file.
Schedule a browser task to run automatically on a fixed interval (min 1 hour). Each run is a
| Parameter | Type | Required | Notes |
|---|---|---|---|
| task | string | yes | Plain-English task instructions (same as run_task). |
| interval_seconds | integer | yes | Run interval in seconds. 3600 (1h) – 2592000 (30d). |
| name | string | no | Optional label for the schedule. |
| llm | string | no | — |
| reasoning | no | — | |
| llm_key_id | string | no | Id of a saved LLM key (see list_llm_keys). Optional. |
| llm_api_key | string | no | Per-schedule LLM key. Rarely needed: prefer a saved key. Overrides llm_key_id. |
| country | string | no | 2-letter egress country. |
| max_steps | integer | no | Max agent steps per run. |
| initial_url | string | no | Optional pre-navigation URL. |
List this workspace's scheduled tasks (id, task, interval, next run, run count). Secrets redacted.
No parameter schema on file.
List the LLM provider keys saved in this workspace: id, name, provider, masked key (prefix + last 4),
No parameter schema on file.
List the hosted tiers (browsa, browsa-fast, browsa-smart: the model each runs on, its default reasoning
No parameter schema on file.
List the output files a task wrote — extracted data + a session.gif recording. Returns name + size per file.
| Parameter | Type | Required | Notes |
|---|---|---|---|
| job_id | string | yes | The job id from run_task / create_task. |
- Every tool, no call limit
- No card, and mcprush charges nothing for it
- Source published under MIT
- 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 Browsa 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 | moderate | capability blast radius (moderate) — client exposure if the model is manipulated | −3 |
| 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 |
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 (20 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. moderate is the level these add up to.
| Tool | Capability tags | Why the tag was assigned |
|---|---|---|
| run_task | no tags | |
| get_job | untrusted-input | untrusted-input: keyword "fetch" |
| cancel_job | no tags | |
| respond_to_job | no tags | |
| start_browser | no tags | |
| proxy_network | no tags | |
| stop_browser | no tags | |
| observe | no tags | |
| act | no tags | |
| list_browsers | no tags | |
| credits | no tags | |
| list_webhooks | no tags | |
| create_webhook | external-sink | external-sink: keyword "webhook" |
| delete_webhook | external-sink | external-sink: keyword "webhook", parameter "webhook" |
| list_identities | no tags | |
| create_schedule | no tags | |
| list_schedules | no tags | |
| list_llm_keys | no tags | |
| list_models | no tags | |
| list_job_files | 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 0.3.3. Pin a release below and the command asks for exactly that one.
No release note was published with this version.
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 Browsa 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.