Live crypto readings, inside your AI

What it gives your assistant

Seven read-only tools over the same data as the site. Answers are short markdown tables with the time they were read and the caveat that a window reading is not a forecast.

get_overview

Every coin in one table: spot, the overall Buy/Sell/Hold call with score and confidence, the 1 hour window and the daily regime call.

get_coin_signal

One coin in full: each window running now and the next, every venue's own reading, the daily regime call and the live backtest. Optional timeframe.

get_regime

The daily regime (trend) board, or one coin's call, how far it is from its 50-day average and the close that would change it.

get_results

The live backtest with fees: one strategy across coins, or all eight strategies for one coin, next to buy and hold.

get_history

One coin's blended reading over the last 1 to 48 hours.

list_coins_and_venues

The coins with venue readings, the windows, and what each venue covers.

get_methodology

Sections of the methodology: what the score is, confidence, options readings, the backtest and the regime rule.

Things to ask once it is connected:

• “What is the BTC signal on PolySignals right now, and how confident is it?”
• “Which coins are long on the daily regime rule, and which are closest to flipping?”
• “How has following the 1 hour calls done for ETH after fees, compared with buy and hold?”
• “Do Polymarket and Kalshi agree on SOL’s 15 minute window?”

Claude

claude.ai, Claude Desktop and the mobile apps share connectors through your account. On the Free plan you can add one custom connector.

1. Open Customize, then Connectors.
2. Choose Add, then Add custom connector.
3. Name it PolySignals and paste https://polysignals.net/api/mcp as the URL.
4. Pick No sign in as the authentication and save.

On Team and Enterprise plans an owner adds it once under Organization settings, Connectors.

Claude Code, in a terminal:

claude mcp add --transport http polysignals https://polysignals.net/api/mcp

Put --scope user before the name to have it in every project, or add it to the project’s .mcp.json:

{
  "mcpServers": {
    "polysignals": {
      "type": "http",
      "url": "https://polysignals.net/api/mcp"
    }
  }
}

ChatGPT

1. In ChatGPT, open the plugins page (chatgpt.com/plugins) and choose the + button.
2. Choose Add custom MCP server.
3. Paste https://polysignals.net/api/mcp as the public endpoint URL and choose no authentication.
4. Create it as a plugin, then turn it on in a chat.

Which ChatGPT plans can add custom MCP servers changes over time; if the option is missing, your plan or workspace settings do not offer it yet.

Cursor

Add to Cursor

Or add it to ~/.cursor/mcp.json (every project) or .cursor/mcp.json (one project):

{
  "mcpServers": {
    "polysignals": {
      "url": "https://polysignals.net/api/mcp"
    }
  }
}

VS Code and GitHub Copilot

Add to VS Code

Or add .vscode/mcp.json to a workspace, then use it from Copilot Chat in agent mode:

{
  "servers": {
    "polysignals": {
      "type": "http",
      "url": "https://polysignals.net/api/mcp"
    }
  }
}

Gemini CLI

gemini mcp add --transport http polysignals https://polysignals.net/api/mcp

Or in settings.json (httpUrl is the streamable HTTP transport; url would mean the older SSE one):

{
  "mcpServers": {
    "polysignals": {
      "httpUrl": "https://polysignals.net/api/mcp"
    }
  }
}

OpenAI Codex

codex mcp add polysignals --url https://polysignals.net/api/mcp

Or in ~/.codex/config.toml:

[mcp_servers.polysignals]
url = "https://polysignals.net/api/mcp"

Windsurf, Zed, Cline and LM Studio

Windsurf (Devin Desktop), in mcp_config.json:

{
  "mcpServers": {
    "polysignals": {
      "serverUrl": "https://polysignals.net/api/mcp"
    }
  }
}

Zed, in its settings:

{
  "context_servers": {
    "polysignals": {
      "url": "https://polysignals.net/api/mcp"
    }
  }
}

Cline, in its MCP settings (without type Cline assumes the older SSE transport):

{
  "mcpServers": {
    "polysignals": {
      "url": "https://polysignals.net/api/mcp",
      "type": "streamableHttp"
    }
  }
}

LM Studio, in mcp.json:

{
  "mcpServers": {
    "polysignals": {
      "url": "https://polysignals.net/api/mcp"
    }
  }
}

Copilot Studio, Le Chat and Perplexity

Microsoft Copilot Studio: in your agent, open Tools, choose Add a tool, then New tool and Model Context Protocol. Give it a name and a description, paste https://polysignals.net/api/mcp and set authentication to None.

Mistral Le Chat: open Connectors, choose Add Connector and the Custom MCP Connector tab, and give it a name without spaces and the URL. In an organisation an admin may have to add it.

Perplexity: plans with custom connectors add a remote one under Settings, Connectors. Whether it accepts a server without sign-in may depend on your plan.

From the OpenAI and Anthropic APIs

Both APIs call a remote MCP server for you. OpenAI’s Responses API, with your own model name:

curl https://api.openai.com/v1/responses \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"YOUR_MODEL","tools":[{"type":"mcp","server_label":"polysignals","server_description":"Live crypto Buy/Sell/Hold readings from prediction markets and options","server_url":"https://polysignals.net/api/mcp","require_approval":"never"}],"input":"What is the BTC signal on PolySignals right now?"}'

Anthropic’s Messages API, with the MCP connector beta header:

curl https://api.anthropic.com/v1/messages \
  -H "x-api-key: $ANTHROPIC_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "anthropic-beta: mcp-client-2025-11-20" \
  -H "content-type: application/json" \
  -d '{"model":"claude-opus-5-5","max_tokens":1024,"mcp_servers":[{"type":"url","url":"https://polysignals.net/api/mcp","name":"polysignals"}],"tools":[{"type":"mcp_toolset","mcp_server_name":"polysignals"}],"messages":[{"role":"user","content":"What is the BTC signal on PolySignals right now?"}]}'

Clients that only run local servers

A client that can only start a local program can reach the server through the mcp-remote bridge (needs Node.js):

{
  "mcpServers": {
    "polysignals": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://polysignals.net/api/mcp"
      ]
    }
  }
}

Without MCP

The same readings are plain markdown at /api/v1/live/index.md (every coin) and /api/v1/live/BTC.md (one coin), which any assistant that can open a URL can read, and as JSON through the API. The server card is at https://polysignals.net/api/mcp/server-card.

Limits

The server is read-only and keeps no state between calls. Up to 50 requests per 10 seconds per IP across the API and this server; above that, or when the server is busy, it answers 429 and a client should wait about 10 seconds. Data refreshes once a minute or faster, so polling faster than that returns the same numbers. Use is subject to the terms.