jev-browser
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| TEXT_MODEL | No | Text model id for the text helper. | inception/mercury-2.5 |
| TYPESAFE_MODEL | No | TypeSafe model id. | jev-latest |
| TYPESAFE_API_KEY | No | TypeSafe Jev API key used for operation + target selection (only needed for the standalone demo/library loop; the MCP server calls no model and requires no keys). | |
| TEXT_MODEL_API_KEY | No | API key for the text helper used only by the TYPE_TEXT action. Works with any OpenAI-compatible provider. | |
| TYPESAFE_DEMO_PORT | No | Port for the local demo inspector. | 8766 |
| TEXT_MODEL_BASE_URL | No | OpenAI-compatible endpoint for the text helper. | https://openrouter.ai/api/v1 |
| TEXT_MODEL_REASONING | No | Set to `none` to disable reasoning in the text helper. | none |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| jev_navigateB | Open a URL in the browser. Creates a new browser session if none exists. Args: url: URL to navigate to |
| jev_observeA | Read current page state: URL, title, text, and interactive elements with IDs. Args: screenshot: Include screenshot (base64 JPEG). Default: false |
| jev_actA | Execute an action on an element. Use action_id from jev_observe. Args: action_id: Element ID from jev_observe (e.g., 'e1', 'e2') text: Text to type (required for fill actions) |
| jev_closeB | Close the browser session. |
| jev_configB | Set or update API configuration at runtime. Args: typesafe_api_key: TypeSafe API key for operation/target selection typesafe_base_url: TypeSafe API base URL (default: https://api.typesafe.ai/v1) text_model_api_key: Text model API key for TYPE_TEXT operations text_model_base_url: Text model API base URL (default: https://api.deepseek.com/v1) text_model: Text model name (default: deepseek-chat) |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 5 tools
Each tool maps to a distinct phase of browser control: observe reads state, navigate opens URLs, act performs element actions, close ends the session, and config handles configuration. The only mild overlap is that act can perform many action types, but it remains clearly separated from the read-only observe.
All five tools follow an identical 'jev_' prefix plus a concise snake_case verb pattern (observe, navigate, act, close, config), making the set highly predictable and readable.
Five tools is lean but well-scoped for a browser automation agent, with each tool earning its place in the observe-navigate-act-close loop. The addition of config is justified for runtime setup.
The core browser lifecycle (navigate, observe, act, close) is covered, and act likely encompasses clicks, typing, and scrolling via element IDs. Minor gaps like explicit back/forward navigation, tab management, or scroll-only primitives exist but agents can work around them.