npm package discovery and stats viewer.

Discover Tips

  • General search

    [free text search, go nuts!]

  • Package details

    pkg:[package-name]

  • User packages

    @[username]

Sponsor

Optimize Toolset

I’ve always been into building performant and accessible sites, but lately I’ve been taking it extremely seriously. So much so that I’ve been building a tool to help me optimize and monitor the sites that I build to make sure that I’m making an attempt to offer the best experience to those who visit them. If you’re into performant, accessible and SEO friendly sites, you might like it too! You can check it out at Optimize Toolset.

About

Hi, 👋, I’m Ryan Hefner  and I built this site for me, and you! The goal of this site was to provide an easy way for me to check the stats on my npm packages, both for prioritizing issues and updates, and to give me a little kick in the pants to keep up on stuff.

As I was building it, I realized that I was actually using the tool to build the tool, and figured I might as well put this out there and hopefully others will find it to be a fast and useful way to search and browse npm packages as I have.

If you’re interested in other things I’m working on, follow me on Twitter or check out the open source projects I’ve been publishing on GitHub.

I am also working on a Twitter bot for this site to tweet the most popular, newest, random packages from npm. Please follow that account now and it will start sending out packages soon–ish.

Open Software & Tools

This site wouldn’t be possible without the immense generosity and tireless efforts from the people who make contributions to the world and share their work via open source initiatives. Thank you 🙏

© 2026 – Pkg Stats / Ryan Hefner

predictable-mcp

v0.2.0

Published

MCP server for prediction market intelligence — unified market data + ICM convergence + cross-exchange divergence alerts across 7 exchanges (Polymarket, Kalshi, Limitless, Probable, Baozi, Myriad, Opinion). 8 tools for AI agents.

Readme

predictable-mcp

MCP server for Predictable — unified prediction market data and ICM convergence signals across 7+ exchanges (Polymarket, Kalshi, Manifold, Metaculus, Limitless, Probable, Myriad).

Primary surface for AI agents (Claude Desktop, Cursor, Cline, ChatGPT Apps, custom MCP clients). The REST API at api.predictable.dev is the secondary surface for non-agent customers.

Status: v0.2.0 — 8 tools with live handlers wired to @eu-predictions/odds-aggregator (pmxt) + @eu-predictions/markets (ICM convergence). Premium tools (get_brier_leaderboard, backtest_signal) gate on PREDICTABLE_API_KEY; the Brier leaderboard's historical DB lands in v0.3.

Transport

This server uses Streamable HTTP (the MCP 2026 transport). The older SSE transport is deprecated.

Tools

| Tool | Tier | Description | |---|---|---| | list_markets | Free | Search and list markets across 7 exchanges, cursor paginated | | get_market | Free | Full market detail, optional orderbook + history | | get_calibration_score | Free | ICM convergence + component breakdown across exchanges | | get_convergence_alerts | Free | Cross-exchange divergence alerts above a basis-point threshold (the moat) | | get_event_weighted_consensus | Free | Volume-/liquidity-/equal-weighted cross-exchange consensus probability | | map_news_to_markets | Free | Map a headline to the top-K affected markets via term overlap | | get_brier_leaderboard | Premium | Historical Brier calibration ranking by exchange × category (requires PREDICTABLE_API_KEY) | | backtest_signal | Premium | Replay momentum / mean-reversion / threshold strategies on OHLCV (requires PREDICTABLE_API_KEY) |

All tools declare outputSchema so hosts can render structured UI, and carry readOnlyHint: true / idempotentHint: true annotations.

TOS safety: every tool is an intelligence-layer derivation (convergence, alerts, consensus, Brier, backtests). We do not redistribute raw exchange orderbooks or resolution feeds, consistent with Kalshi Data TOS and similar.

Install (end user)

Claude Desktop / Cursor / Cline

Add to your MCP client config (e.g. claude_desktop_config.json):

{
  "mcpServers": {
    "predictable": {
      "type": "url",
      "url": "https://mcp.predictable.dev/mcp",
      "headers": {
        "Authorization": "Bearer <YOUR_PREDICTABLE_API_KEY>"
      }
    }
  }
}

For local development:

{
  "mcpServers": {
    "predictable": {
      "type": "url",
      "url": "http://127.0.0.1:3333/mcp"
    }
  }
}

npm (once published)

npx predictable-mcp

Develop (this repo)

pnpm install
pnpm --filter predictable-mcp dev        # tsx watch mode
pnpm --filter predictable-mcp build      # emit dist/
pnpm --filter predictable-mcp typecheck  # strict TS check
pnpm --filter predictable-mcp inspect    # MCP inspector UI

Configuration (env vars)

| Variable | Default | Purpose | |---|---|---| | PREDICTABLE_API_URL | https://api.predictable.dev/v1 | Backend REST base URL | | PREDICTABLE_API_KEY | — | Bearer token for authenticated / premium tier | | MCP_HTTP_PORT | 3333 | HTTP listen port | | MCP_HTTP_HOST | 127.0.0.1 | HTTP listen host |

Never commit secrets. Use .env.local (git-ignored) or the host process environment.

Architecture

AI agent (Claude / Cursor / Cline / ChatGPT)
   └── MCP Streamable HTTP  →  predictable-mcp
                                    ├── list_markets           → REST  /v1/markets/list
                                    ├── get_market             → REST  /v1/markets/get
                                    └── get_calibration_score  → REST  /v1/markets/calibration
                                                                     └── packages/icm + convergence-scoring

The MCP server is a thin transport+schema layer. All business logic stays in packages/odds-aggregator (pmxt adapters for 7 exchanges) and packages/icm (convergence engine). The web app, enterprise dashboards, and MCP server all call the same REST backend.

Roadmap

  • v0.1.0 — scaffold: Streamable HTTP transport, 3 tool stubs, input+output schemas, annotations
  • v0.2.0 (this release) — 8 tools with live handlers (3 core + 5 intelligence-layer), wired to @eu-predictions/odds-aggregator (pmxt) + @eu-predictions/markets (ICM convergence). Premium tools gate on PREDICTABLE_API_KEY.
  • v0.3.0 — historical resolution feed for get_brier_leaderboard, audit logging, rate limiting
  • v0.4.0 — Smithery + Glama listings, install guides for Cline / ChatGPT Apps
  • v0.5.0 — x402 pay-per-call rail (reuse OraClaw pattern), publish to npm + MCP registry
  • v1.0.0 — Enterprise tier with SLA, SSO, dedicated rate limits

Publish (once v0.5+)

  1. pnpm --filter predictable-mcp build
  2. npm publish (public or org scope)
  3. mcp-publisher login && mcp-publisher publish (JWT expires ~24h, repeat each release)
  4. Submit to the 5+ owned awesome-mcp-* directories (zero-budget distribution)

License

UNLICENSED — private source. Public distribution TBD at v1.0.