mcpField
B
Bitview

Bitview

Remote
mcp.bitview.spaceFinance177Updated 5w ago

About Bitview

Bitview is a read-only MCP server for Bitcoin blockchain, mempool, mining, market and on-chain analytics, requiring no API key. It is a stateless adapter that forwards tool calls as GET requests to the Bitview REST API.

Key features

  1. 01Read-only blockchain and mempool data
  2. 02Mining and market data queries
  3. 03On-chain analytics metrics
  4. 04No authentication required
  5. 05Tool catalog generated from OpenAPI

Use cases

  1. 01Looking up Bitcoin block and transaction information
  2. 02Monitoring mempool and mining status
  3. 03Analyzing on-chain data and market indicators

MCP support

The server exposes the Bitview REST API operations as generated MCP tools over a Streamable HTTP endpoint and forwards each call to the upstream REST API.

Configuration

Remote servers need no local install. Pick your client, then copy the config or install in one click.

Endpointhttps://mcp.bitview.space/
Copy install command
Claude CodeTerminal
Cursor.cursor/mcp.json
VS Code.vscode/mcp.json
Claude Desktopclaude_desktop_config.json
Windsurf~/.codeium/windsurf/mcp_config.json

Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.

Tools

Details

From the project README

bitview_mcp

bitview_mcp is a thin, stateless, read-only MCP adapter for the Bitview REST API. It exposes the MCP-visible OpenAPI operations as generated tools and forwards every tool call to the configured public REST origin as a GET request.

The official public endpoint is mcp.bitview.space. It is stateless, read-only, and requires no authentication.

A server built from the current source also serves an instance landing page, /privacy, /terms, /support, and /logo.png. The icon is embedded in the binary and advertised in MCP discovery metadata. These instance-local routes are separate from the MCP protocol and may not be available on older deployed versions.

Caching model

bitview_mcp does not cache API responses or retain MCP sessions. Point it at the public Cloudflare-fronted REST origin so its upstream GET requests use the existing Cloudflare cache:

MCP client -> bitview_mcp -> Cloudflare-cached REST API -> Bitview server

Cloudflare does not need to cache the MCP endpoint. The MCP catalog TTL is only a standard client-side cache hint for the static discovery and tool-list metadata.

Run

From the workspace:

cargo run -p bitview_mcp -- \
  --api https://bitview.space \
  --url https://mcp.bitview.space/ \
  --name Bitview

Or run an installed binary:

bitview_mcp \
  --api https://api.example.com \
  --url https://mcp.example.com/ \
  --name "Example Node"

All three options are required:

  • --api is the upstream Bitview REST origin. A bare host tries HTTPS first and retries over HTTP only if HTTPS fails at the transport layer.
  • --url is this server's public MCP URL. It must be an absolute HTTP(S) origin.
  • --name is the human-readable instance name shown on the landing page and in MCP discovery metadata.

For local development:

bitview_mcp \
  --api http://127.0.0.1:3110 \
  --url http://127.0.0.1:3111/ \
  --name "Local Bitview"

The Streamable HTTP endpoint is http://127.0.0.1:3111/ by default. If that port is unavailable, the server tries each port through 3211. The server supports MCP protocol version 2026-07-28.

Both URLs must be origins without a path, query, or credentials. The public URL is normalized with a trailing slash.

Generated catalog

The tool catalog is generated from the canonical OpenAPI document by bitview_bindgen and embedded in the binary at compile time from generated/manifest.json. Do not edit the generated manifest by hand; after changing the API, run:

cargo run -p bitviewd --example bindgen --features bindgen

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.