Browser tasks with MCP

Connect the pre.dev MCP server once. It provides browser_agent, browser_agent_get, and browser_agent_list alongside the specification tools.

Extract structured data

Call browser_agent with these arguments:

{
  "tasks": [{
    "url": "https://example.com",
    "instruction": "Extract the main page heading.",
    "output": {
      "type": "object",
      "properties": { "heading": { "type": "string" } },
      "required": ["heading"]
    },
    "mode": "extract",
    "maxSteps": 10,
    "maxDurationSeconds": 120
  }]
}

The tool returns a text summary and run ID. Call browser_agent_get with that ID to get the complete structuredContent, including results[i].data.

Run a longer workflow

For a multi-step workflow, use mode: "agent", give a concrete completion condition, and submit asynchronously:

{
  "tasks": [{
    "url": "https://news.ycombinator.com/newest",
    "instruction": "Open the next page using More, then collect the first five story titles and links.",
    "successCondition": "The next page has been opened and five stories from that page have been collected.",
    "output": {
      "type": "object",
      "properties": {
        "stories": {
          "type": "array",
          "items": {
            "type": "object",
            "properties": {
              "title": { "type": "string" },
              "url": { "type": "string" }
            },
            "required": ["title", "url"]
          },
          "minItems": 5,
          "maxItems": 5
        }
      },
      "required": ["stories"]
    },
    "mode": "agent",
    "maxSteps": 20,
    "maxDurationSeconds": 180
  }],
  "async": true
}

Save the ID from the returned text, then call browser_agent_get every few seconds. Stop when batch status becomes completed or failed, and inspect each task’s uppercase status. A completed batch does not mean every task succeeded.

Inspect evidence

{ "id": "507f1f77bcf86cd799439011", "includeEvents": true }

This returns task timelines and any available screenshots. During synchronous execution, clients with logging support can also receive notifications/message. These are MCP notification envelopes, not REST SSE frames; heavy event data may be shortened.

History and limits

Use browser_agent_list with limit, skip, and optionally status: "processing" or "completed". Its structured response has total and batches; use browser_agent_get for each run’s full data. The same credit, batch-size, and account in-flight gates apply as in REST. concurrency allows 1–20 tasks within a run, default 5. Retrieve queue status through REST when you need the current account limit. The MCP tool does not expose REST’s idempotency header or a cancellation operation. Closing the client does not establish that browser work stopped. See the complete tool reference for every field and the older browser_task aliases.

​Extract structured data

​Run a longer workflow

​Inspect evidence

​History and limits

Extract structured data

Run a longer workflow

Inspect evidence

History and limits