Quick start examples for the Lexmount Node.js SDK.
- Visit the Lexmount website
- Verify the page title
- Take a screenshot
- Query the public endpoint catalog through
client.catalogInfo() - Print available regions and hosts
- Build a direct websocket URL from
LEXMOUNT_BASE_URL - Connect through
/connection?project_id=...&api_key=... - Visit
https://example.comand saveconnection_demo.png
- Create a browser session with
customImageId - Accept
--custom_image_idfrom the command line - Connect to the session and verify the browser can open a page
- Create a browser session with
windowSize - Accept
--window_size(also--window-sizeor--window size), defaulting to1920,1080 - Connect to the session and print the live initial viewport
- Use
lightbrowser mode - Enable LightMount layout and show its per-session
enableLightmountResourceswitch - Visit Sina News
- Extract all links and save them to
links.txt
- Create test sessions
- List sessions with pagination information
- Filter sessions by status
- Clean up sessions
- Create a context with
description - Start a
readWritesession with that context
- Create contexts with descriptions
- List contexts and print display names
- Get details for a specific context, including its display name
- Clean up created contexts
- Accept an existing source
context_id - Fork it into a new context
- Print the forked context id
- Create a read-write context session
- Demonstrate lock conflict handling through
ContextLockedError
- Create a context
- Run one
readWritesession and two concurrentreadOnlysessions
- Upload a browser extension archive
- List uploaded extensions
- Create a session with
extensionIds
- List uploaded extensions
- Get details for one extension
- Optionally upload and delete an extension when
LEXMOUNT_EXTENSION_PATHis set
- Create a session with
proxy - Verify the remote browser can access pages through the upstream proxy
- Access an internal page using the SDK machine's network and DNS
- Open a tunnel, navigate and screenshot, then clean up the session and tunnel
- Create a session with
officialProxy: true - Verify the remote browser can access pages through the Lexmount official proxy pool
- Create a browser session
- Print the
inspectUrlfor manual inspection - Wait for user input before closing the session
- Create a browser session
- Query
/jsontargets through the SDK - Print each target's
inspectUrl, page URL, and websocket URL
- Trigger a remote file download
- Query session downloads through the SDK
- Archive session downloads to a local zip file
Requires Node.js 18.17 or newer for the npm run commands below (which use tsx).
You can also run node demo.ts directly with Node.js 22.18+ or 24+; the same applies to every demo. See Node.js TypeScript support.
# 1. Install dependencies
npm install
# 2. Create .env from the template
cp .env.example .env
# On a local macOS/Windows terminal, missing credentials trigger browser sign-in.
# 3. Run examples
npm run demo
npm run catalog-info
npm run connection-demo
npm run custom-image-demo -- --custom_image_id code.lexmount.net/neng/chrome:tag
npm run window-size-demo -- --window_size 800,600
npm run light-demo
npm run session-list
npm run context-basic
npm run context-list-get
npm run context-fork -- <context_id>
npm run context-lock-handling
npm run context-modes
npm run extension-basic
npm run extension-list-get
npm run proxy-demo
npm run local-proxy-demo -- --url http://oa.company.internal/
npm run official-proxy-demo
npm run inspect-url-demo
npm run session-targets
npm run session-downloadsThe .env file should contain:
LEXMOUNT_API_KEY=your_api_key_here
LEXMOUNT_PROJECT_ID=your_project_id_here
LEXMOUNT_BASE_URL=https://api.lexmount.com
LEXMOUNT_EXTENSION_PATH=/absolute/path/to/extension.zip
LEXMOUNT_PROXY_SERVER=http://host:port
LEXMOUNT_PROXY_USERNAME=
LEXMOUNT_PROXY_PASSWORD=
LEXMOUNT_CUSTOM_IMAGE_ID=code.lexmount.net/neng/chrome:tag
LEXMOUNT_WINDOW_SIZE=1920,1080Every demo checks LEXMOUNT_PROJECT_ID and LEXMOUNT_API_KEY before making API requests. It loads .env from the current working directory, with .env taking precedence over exported environment variables. Empty values and the example placeholders count as missing. Existing complete credentials are used without opening a browser.
- Default API:
https://api.lexmount.com; website:https://browser.lexmount.com. - With missing credentials, an interactive local macOS or Windows terminal opens the system browser for sign-in and authorization. Return to the terminal after approving; the demo continues automatically.
- The PKCE flow uses a temporary
127.0.0.1callback and exchanges a one-time code over HTTPS. Both credentials and the matching API base URL are saved together to.env, preserving unrelated settings. Newly written files have owner-only permissions on POSIX systems; on Windows protect the project directory with your user account's ACLs. - On Linux, SSH, CI, non-interactive terminals, or when browser authorization fails/times out (3 minutes), the demo exits with website/manual setup instructions. Fill in the two values and rerun. You can also set both environment variables without an
.envfile in CI. - Explicit
https://api.lexmount.cnuseshttps://browser.lexmount.cnfor authorization. Other custom API URLs are preserved and require manual credentials from their matching environment; they never silently log in to.com. - No credentials are saved on an unsuccessful exchange. If
.envis changed while signing in, rerun the demo to avoid overwriting those edits. Do not commit.env.
Run examples from this repository directory so they share the same .env.
local-proxy-demo.ts uses SDK 0.7.0, opens a separate local-network tunnel, then creates a normal cloud browser with that tunnel. Run it on the machine connected to your company network or VPN. Target DNS and TCP connections use this machine's network. The API environment and selected region must have the local proxy gateway deployed, and the project must have custom_proxy enabled.
After installing dependencies and setting credentials, run:
npm run local-proxy-demo -- --url http://oa.company.internal/
# Optional: select a catalog region available at LEXMOUNT_BASE_URL
npm run local-proxy-demo -- --url http://oa.company.internal/ --region <region-id>The target URL is required via --url. You can set optional LEXMOUNT_REGION in .env; --region takes precedence. --help works without credentials or API access.
The demo prints the page title, saves local_proxy_demo.png, then waits for any key before closing the browser, cloud session, tunnel and client. The browser and local connector stay available while waiting; Ctrl+C also triggers cleanup. End-of-input allows noninteractive runs to finish. Sessions and tunnels are also closed when navigation fails. One client keeps the project, API key and region consistent.
Use an internal hostname or LAN IP for the target. Chrome has implicit proxy bypass rules for localhost/loopback and link-local addresses, so 127.0.0.1 is not a reliable address for this demo. The cloud browser still needs its own website login and certificate trust.
This repository pins Lexmount SDK 0.7.0, published on npm. Run npm install to install it; a sibling SDK checkout is not required. The server must still have its public Browser Agent API enabled and the project configured for access.
All new examples use the credential setup described above: existing credentials are reused; on local macOS/Windows terminals, missing credentials start browser sign-in and authorization, then save the returned Project ID and API key.
# Use the automatic sign-in flow above, or configure .env beforehand.
export LEXMOUNT_BASE_URL='https://api.lexmount.com'
npm run browser-agent
npm run browser-agent-advanced
npm run browser-agent-recoverybrowser-agent.ts: create a standard Run, then wait; 201 means accepted. Task failure is a normal Run result.browser-agent-advanced.ts: save an Agent with a JSON Schema and{{topic}}variable; read native object output and usage. The saved Agent remains available for reuse.browser-agent-recovery.ts: persist a business ID, exact request, idempotency key and Run ID in a local file before/after POST. Rerun the same file to recover. A lost response uses the original key; a known Run uses GET/wait. Do not use an unknown receipt's key beyond 7 days without reconciliation.
Set RUN_IDEMPOTENCY_KEY to reuse a Run create across retries, and a separate AGENT_IDEMPOTENCY_KEY for Agent creation. Repeating the advanced example requires both keys so the Agent ID in the Run request stays unchanged. Prefer the recovery example for process restarts. Separate independent creates need separate keys. SDK retries within one method call reuse its key.
RunWaitTimeout only stops local waiting; it does not cancel or recreate the Run. Resume GET/wait with the saved ID. Keep API keys out of URLs, logs and public browser code. usage.credits=null means unconfirmed metering; browser time is billed separately. The SDK and these examples default to https://api.lexmount.com; an explicit LEXMOUNT_BASE_URL still selects CN or a custom environment.
These examples use the existing credential setup flow described above.
Run bash scripts/test-browser-agent.sh in a clean checkout. It installs the declared 0.7.0 package from npm and checks credential setup, recovery and TypeScript types without a sibling SDK checkout. These checks use local mocks and do not create real Agents or Runs.
Recovery files bind the original Project ID and API origin, with no API key saved. Reusing a file in another project/service, using an old unscoped file or losing its idempotency key is rejected before any Run request. Reconcile the original task instead of deleting the file and rerunning. On POSIX, file contents and directory entries are synced before POST and after the receipt rename. A known Run ID uses GET/wait even after the seven-day create-key window.