Algora MCP Server

Algora MCP Server

An MCP (Model Context Protocol) server that gives AI agents direct access to Algora's open-source bounty platform. Use it from Claude Desktop, Cursor, Windsurf, or any MCP-compatible client to discover, search, and analyze open bounties.

What It Does

AI agents can call five tools:

Requirements

• Node.js 18+
• No API key required — Algora's public API is open

Quick Start

git clone https://github.com/idapixl/algora-mcp-server
cd algora-mcp-server
npm install
npm run build

Claude Desktop Config

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "algora": {
      "command": "node",
      "args": ["/absolute/path/to/algora-mcp-server/dist/index.js"]
    }
  }
}

Cursor / Windsurf Config

Add to your MCP settings:

{
  "algora": {
    "command": "node",
    "args": ["/absolute/path/to/algora-mcp-server/dist/index.js"]
  }
}

Example Agent Interactions

Find high-value TypeScript bounties:

"Show me open bounties worth $200+ in TypeScript"

The agent will call get_top_bounties({ tech: "typescript", min_amount: 200 }) and return formatted results.

Check a specific org:

"What bounties does projectdiscovery have open?"

Calls get_org_bounties({ org: "projectdiscovery" }).

Search by domain:

"Find bounties related to MCP or model context protocol"

Calls search_bounties({ keyword: "MCP" }).

Market overview:

"What's the total value of all open bounties on Algora?"

Calls get_bounty_stats({}) and returns counts, total, average, breakdown by tech and org.

Tool Reference

list_bounties

get_org_bounties

search_bounties

get_top_bounties

get_bounty_stats

Development

npm run dev      # Run with tsx (no build step)
npm run build    # Compile TypeScript
npm run typecheck  # Type check without emitting

Architecture

The server uses stdio transport (standard MCP convention) — no HTTP server, no port, no configuration. It communicates with the MCP client through stdin/stdout and connects to Algora's public API (no auth required).

The @algora/sdk tRPC client handles API calls. Local filtering is applied in-memory for features the API doesn't natively support (keyword search, min/max amount).

License

MIT