An MCP (Model Context Protocol) server that gives AI assistants the ability to analyze blockchain transactions across EVM, Solana, and TON networks.
Connect it to Claude, Cursor, Windsurf, or any MCP-compatible client and start debugging transactions in natural language — no API keys required on your end.
| Ecosystem | Networks |
|---|---|
| EVM | Ethereum, Polygon, Arbitrum, Optimism, Base, BSC |
| Solana | Mainnet, Devnet |
| TON | Mainnet, Testnet |
| Tool | Description |
|---|---|
debug_transaction |
Full pipeline analysis: call tree, token flows, semantic actions, risk flags, failure reasons, and an AI narrative. Supports EVM, Solana, and TON. |
resolve_rango_swap |
Resolve a Rango cross-chain swap by swap ID — returns all steps, statuses, and per-chain transaction hashes. |
ask_question |
Ask a follow-up question about a previously analyzed transaction. |
Skills are pre-built prompt templates that guide the AI through complex workflows. Invoke them from your MCP client's slash menu or prompt library.
| Skill | Use When |
|---|---|
debug_evm_transaction |
Analyzing any EVM transaction — swap, bridge, NFT mint, contract call |
debug_solana_transaction |
Analyzing Solana transactions — SPL transfers, Jupiter swaps, staking |
debug_ton_transaction |
Analyzing TON transactions — Jetton transfers, DeDust/STON.fi swaps |
trace_rango_swap |
Tracing a complete Rango cross-chain swap end-to-end across all chain legs |
investigate_failed_transaction |
Root-cause analysis of a failed transaction, with plain-English explanation |
Each skill follows a structured prompt format covering task context, tone, background data, rules, examples, step-by-step reasoning instructions, output formatting, and a prefilled response opener.
If you have been given a server URL and API key, add the following to your MCP client configuration and you're ready to go — no local setup required.
Claude Desktop (~/Library/Application Support/Claude/claude_desktop_config.json on macOS):
{
"mcpServers": {
"rango-debugger": {
"url": "https://your-mcp-server-url/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}Cursor / Windsurf — add the same entry under MCP settings in your IDE preferences.
- Node.js 20+
- A running instance of the Rango Transaction Debugger backend (or use the hosted backend)
Suitable for use with Claude Desktop or Cursor when running everything on your own machine.
git clone https://github.com/0x4rde/rango-transaction-debugger.git
cd rango-transaction-debugger
npm install
cp .env.example .env
# Edit .env: set BACKEND_URL to your backend instance
npm run buildThen add to your MCP client config:
{
"mcpServers": {
"rango-debugger": {
"command": "node",
"args": ["/absolute/path/to/rango-transaction-debugger/dist/index.js"]
}
}
}Run the HTTP server so your whole team can connect without local setup.
cp .env.example .env
# Edit .env:
# BACKEND_URL=https://your-backend-url
# MCP_API_KEY=<generate with: openssl rand -hex 32>
# PORT=3100
npm run build
npm run start:serverOr with Docker:
docker compose up -dThe MCP endpoint will be available at http://localhost:3100/mcp.
Put it behind a reverse proxy (nginx, Traefik, Caddy) for HTTPS.
| Variable | Default | Description |
|---|---|---|
BACKEND_URL |
(required) | URL of the Rango Transaction Debugger backend API |
MCP_API_KEY |
(none) | Bearer token for authentication. Required when exposing the server over the internet. |
PORT |
3100 |
HTTP port for the server entry point |
Once connected, you can talk to your AI assistant naturally:
Debug this transaction: 0xabc123...
Use the trace_rango_swap skill for swap ID: a1b2c3d4-...
Why did this transaction fail? 0xdef456...
What tokens were transferred in 0x789abc... on Arbitrum?
MIT