mcpField
W
Web Task

Web Task

Local
github.com/jojo-labsBrowser Automation
Repository

About Web Task

Browser tasks, sessions, and recipes for MCP and HTTP automation

Configuration

Local server config is the same for every client. Copy it into your client’s MCP config file.

mcp.json · Universal config
{
  "mcpServers": {
    "jojo-labs-web-task-api": {
      "command": "npx",
      "args": [
        "-y",
        "web-task-api"
      ],
      "env": {
        "WEB_TASK_API_DATA_DIR": "<WEB_TASK_API_DATA_DIR>",
        "WEB_TASK_API_RECIPES_DIR": "<WEB_TASK_API_RECIPES_DIR>",
        "WEB_TASK_API_TEMP_DIR": "<WEB_TASK_API_TEMP_DIR>",
        "WEB_TASK_API_MODEL": "<WEB_TASK_API_MODEL>",
        "WEB_TASK_API_VARIANT": "<WEB_TASK_API_VARIANT>",
        "WEB_TASK_API_PI_URL": "<WEB_TASK_API_PI_URL>",
        "WEB_TASK_API_PI_MODEL": "<WEB_TASK_API_PI_MODEL>",
        "WEB_TASK_API_PI_VARIANT": "<WEB_TASK_API_PI_VARIANT>",
        "CLIPROXY_BASE_URL": "<CLIPROXY_BASE_URL>",
        "CLIPROXY_AUTH_TOKEN": "<CLIPROXY_AUTH_TOKEN>",
        "CLIPROXY_MODEL": "<CLIPROXY_MODEL>",
        "OPENCODE_BASE_URL": "<OPENCODE_BASE_URL>",
        "OPENCODE_MODEL": "<OPENCODE_MODEL>",
        "BROWSER_USER_DATA_DIR": "<BROWSER_USER_DATA_DIR>",
        "BROWSER_CHANNEL": "<BROWSER_CHANNEL>",
        "HEADLESS": "<HEADLESS>"
      }
    }
  }
}
Run commandnpx -y web-task-api
Environment variables to set
WEB_TASK_API_DATA_DIROptional mutable data root for web-task-api runs, sessions, profiles, and temp state. When omitted, the runtime uses a platform user-data directory instead of the ambient cwd.
WEB_TASK_API_RECIPES_DIROptional recipes directory override. When omitted, web-task-api reads bundled recipes from the installed package.
WEB_TASK_API_TEMP_DIROptional temp root override used when incoming temp env variables point at your home directory or working directory.
WEB_TASK_API_MODELOptional shared default planner model for catalog-backed PI and OpenCode selections. Defaults to deepseek/deepseek-v4-flash.
WEB_TASK_API_VARIANTOptional shared default planner thinking variant for catalog-backed PI and OpenCode selections. Defaults to high.
WEB_TASK_API_PI_URLOptional PI engine base URL for the preferred planner backend. Defaults to http://127.0.0.1:8793.
WEB_TASK_API_PI_MODELOptional PI-specific planner model override. Defaults to WEB_TASK_API_MODEL when omitted.
WEB_TASK_API_PI_VARIANTOptional PI-specific thinking variant override. Defaults to WEB_TASK_API_VARIANT when omitted.
CLIPROXY_BASE_URLOptional CLIProxy base URL for the legacy fallback planner backend.
CLIPROXY_AUTH_TOKENSecretOptional CLIProxy auth token for the legacy fallback planner backend.
CLIPROXY_MODELOptional CLIProxy model alias for the legacy fallback planner backend.
OPENCODE_BASE_URLOptional OpenCode base URL for the compatibility planner backend.
OPENCODE_MODELOptional OpenCode model name for the compatibility planner backend. Defaults to WEB_TASK_API_MODEL when omitted.
BROWSER_USER_DATA_DIROptional existing Chromium/Chrome user-data directory for warmed browser state reuse.
BROWSER_CHANNELOptional Playwright browser channel override, for example `chrome`.
HEADLESSOptional browser headless toggle. Set `false` to prefer a visible browser.

Local servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.

Packages
npmweb-task-api0.10.1

Tools

11

Saved tool list · updated 2w ago

ParameterTypeRequiredDescription
goalstringYesWhat the browser task should accomplish right now.
modeenumNoTask mode: read for extraction or research, act for interactive changes.read | act
inputobjectNoStructured task inputs that the recipe or planner can reference.
profilestringNoOptional named browser profile for authenticated or warmed browser state.
headlessbooleanNoWhether to prefer headless browser mode for this run.
max_stepsintegerNoMaximum browser steps before the run stops.
recipe_idstringNoOptional starter recipe id when you want a known flow instead of a purely freeform run.
start_urlstringNoOptional starting URL when you already know where the task should begin.
agent_kindenumNoPlanner backend to use for this run.auto | pi | cliproxy | opencode | mock
session_idstringNoOptional saved session to reuse when this task should continue earlier browser state or context.
timeout_msintegerNoOverall run timeout in milliseconds.
agent_modelstringNoOptional planner model override for this run. Auto, PI, and OpenCode use the catalog-backed selection when set. Defaults to deepseek/deepseek-v4-flash.
agent_variantstringNoOptional planner thinking variant override for this run. Defaults to high for catalog-backed planners.
output_schemaobjectNoOptional JSON Schema-like object used to validate the final structured result.

Details

The maintainer has not added details yet. Check the README in the repository.

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.