Skip to content
MCP server

SigRank SignalAF — AI Operator Cascade & Leaderboard

Local token-usage scanner that scores your AI coding efficiency and ranks you on a board

Runs locally
Install
Pricing
Free
Platforms
CLI, Linux, macOS
Runs
Locally
Version
1.0.47
Listing updated
4 days ago

Similar MCP servers

About SigRank SignalAF — AI Operator Cascade & Leaderboard

SigRank SignalAF — AI Operator Cascade & Leaderboard is a free MCP server for analytics and data. It runs locally on your machine and is started by your AI client.

SigRank reads usage metadata from supported local AI session logs on your machine and shows input, output, cache-write and cache-read counts over several time windows. From these it calculates Yield (cache_read x output / input squared), Leverage, Velocity, SNR and 10xDEV locally. The authors state it measures observable token patterns, not model quality, work quality or productivity.

Nothing is published unless you choose to. Enrolment binds a device signing key to your account, and a dry-run shows the payload before submit_verified sends only the four token counts, signed. The server re-scores submissions, so a local preview is not a guaranteed public rank. The package needs Node.js 20 or later and also runs as a terminal dashboard with npx sigrank.

MCP tools split into local inspection (tokenpull, rank_windows, tokenpull_compare), calculating from supplied counts, reading public results, exploring what-if changes and enrolling or submitting. An optional loopback proxy exists for clients that do not store provider usage.

Install

Run in a terminal

claude mcp add sigrank-mcp -- npx -y sigrank

Add to claude_desktop_config.json

{
  "mcpServers": {
    "sigrank-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "sigrank"
      ]
    }
  }
}
Install in CursorOpens Cursor, which asks you to confirm. Or add it by hand:

Add to ~/.cursor/mcp.json

{
  "mcpServers": {
    "sigrank-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "sigrank"
      ]
    }
  }
}
Install in VS CodeOpens VS Code, which asks you to confirm. Or add it by hand:

Add to .vscode/mcp.json

{
  "servers": {
    "sigrank-mcp": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "sigrank"
      ]
    }
  }
}

Add to ~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "sigrank-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "sigrank"
      ]
    }
  }
}

Add to your client's MCP settings (most use this shape)

{
  "mcpServers": {
    "sigrank-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "sigrank"
      ]
    }
  }
}

Using several servers? Build one config file with the MCP config generator.

Tools this server provides (13)

Taken from the project documentation; the running server may differ.

tokenpull
Inspect local usage
rank_windows
Rank across time windows
tokenpull_compare
Compare token counts across local sources
rank_paste
Calculate from supplied counts
get_sigrank_standard_record
Standard record from supplied counts
get_leaderboard
Read the public leaderboard
get_operator
Read an operator's public profile
discover_peers
Find peer operators
diagnose_cascade
Diagnose the token cascade
simulate_change
Simulate a change to usage
suggest_improvements
Suggest improvements
enroll
Bind a device to an account
submit_verified
Publish a signed snapshot

What it does

Local usage scan
Reads token counts from supported AI session logs without leaving the machine.
Yield and related metrics
Computes Yield, Leverage, Velocity, SNR and 10xDEV locally.
Terminal dashboard
Tabs for dashboard, trends, compare, board, watch and connect.
Dry-run publishing
Inspect the signed payload before choosing to submit.
Public leaderboard
Read rankings and operator profiles from the board.
What-if tools
Diagnose the token cascade and simulate changes.

Good for

  • See how efficiently your sessions reuse cached context
  • Compare token counts across local AI clients
  • Preview a signed snapshot before publishing
  • Look up the public leaderboard from an agent
  • Test how a change to your workflow might alter your score

Pricing

Free

Free to run; publishing to the leaderboard needs a signalaf.com account

Prices change. Check signalaf.com before you buy.

SigRank SignalAF — AI Operator Cascade & Leaderboard FAQ

Does my prompt content leave my machine?

The site says no prompt content leaves the machine; only four token counts are submitted, and only when you publish.

What do I need to run it?

Node.js 20 or later. Run npx sigrank; no global install is required.

Is publishing automatic?

No. submit_verified requires confirm: true and a dry run is available.

How is rank computed?

By Yield, defined on the site as cache_read x output / input squared.

Is SigRank SignalAF — AI Operator Cascade & Leaderboard free?

Yes. SigRank SignalAF — AI Operator Cascade & Leaderboard is free to use.

How do I add SigRank SignalAF — AI Operator Cascade & Leaderboard to Claude Code?

Run this command in your terminal: claude mcp add sigrank-mcp -- npx -y sigrank

What are the best alternatives to SigRank SignalAF — AI Operator Cascade & Leaderboard?

The closest alternatives to SigRank SignalAF — AI Operator Cascade & Leaderboard are RankCLI, Kestrel, Agnes, Skill Stats Reporter and Tenhaw Kit.

Where this comes from

  • Description: written from the project's README and the project's website.
  • Install steps: from the official MCP Registry entry.

How AIPoweredBot checks servers