Xcatcher

Xcatcher MCP (Remote) is a cloud-based, high-throughput MCP server that enables agent-driven crawling of fresh and latest posts across large sets of usernames. It provides a streamable HTTP interface for task creation, monitoring, and results delivery, with optional on-chain top-ups via x402 and an OpenAPI spec for easy integration by agent builders.

How to use

You interact with Xcatcher MCP (Remote) through its REST API and optional OpenAPI tooling to create crawl tasks, top up points using x402, poll task status, and download results. Use this flow to efficiently monitor a large set of usernames and retrieve structured results for your pipelines or agents.

How to install

There is no local MCP server installation required for using the Remote MCP. You access the service over HTTPS and authenticate with API keys obtained via the REST API. Start by registering an account, then log in to obtain an API key. Use the API key to authorize requests for task creation, status checks, and downloads.

Endpoints and usage patterns

Core interactions follow a create → poll → download pattern. You first register or log in to obtain an API key, then create a crawl task specifying your target users and mode. You poll for the task status until it completes, and finally download the resulting data in XLSX format. You can also request an x402 quote and perform a top-up when you need more points.

Key endpoints include account management, task lifecycle, and x402 top-ups. Use the provided API base URL for all REST calls, and authenticate every request with your Bearer API key.

OpenAPI and tooling

An OpenAPI spec is available for import into agent builders and API tools. Importing the spec helps you wire up points, quotes, top-ups, and downloads quickly. The Remote MCP design supports tool-style orchestration, enabling you to create tasks, poll until completion, and download results in a structured workflow.

End-to-end example workflow

The following represents typical interactions you perform via REST calls, including creating a task, handling a 402 top-up flow, and downloading results. Use the same idempotency key pattern when retrying after a top-up.

OpenAPI import and discovery

Import the raw OpenAPI specification to quickly wire REST calls for tasks, quotes, top-ups, and downloads. You can then use the OpenAPI-driven tooling to generate client code and integration flows for agent frameworks.

Output format

The default export is XLSX. If you need CSV, convert the downloaded XLSX client-side.

Modes and cost visibility

Default mode is normal, which prioritizes speed and low latency for latest posts monitoring. You can opt for a deep mode for richer per-user collection and historical context. The server reports cost and remaining balance after task creation, so you can plan top-ups accordingly.

Error handling and retries

Common errors include missing or invalid Bearer tokens, 402 top-up requirements, and 5xx transient errors. For rate limiting, respect Retry-After headers and apply exponential backoff for retries.

Security and access control

All MCP and REST calls require an Authorization: Bearer header with your API key. Treat API keys as secrets and avoid exposing them in client-side code or publicly accessible logs.

x402 top-up workflow

If a task requires more points, you may receive a 402 PAYMENT_REQUIRED with a payment quote. You then pay the quoted amount through the supported networks (Base or Solana) and provide a signed top-up proof in the PAYMENT-SIGNATURE header. After the top-up, retry the original task using the same idempotency key.

HTTP endpoints overview

Important REST endpoints include authentication, me (points check), task creation, task status, and download. The 402 top-up flow is exposed via the x402/quote and x402/topup endpoints. Always use the Authorization header with your API key for all requests.

Available tools (programmatic capabilities)

The MCP provides an agent-friendly core with the following tools for orchestration: create_crawl_task, x402_topup, get_task_status, get_result_download_url, and cancel_task. Use these tools to drive your crawl workflow from your agent or automation framework.

Available tools

create_crawl_task

Create a new crawl task that consumes points and starts the data collection process.

x402_topup

Top up the current Bearer key using a PAYMENT-SIGNATURE after proving on-chain payment.

get_task_status

Poll the task status until the crawl completes or fails.

get_result_download_url

Fetch an authenticated URL to download the resulting data file.

cancel_task

Cancel a queued task; refunds may be applied per policy.