SatRank
Lightning Network trust oracle for AI agents. Provides real-time node reachability checks, trust scores, and personalized pathfinding for 17,000+ Lightning nodes via 12 MCP tools.
README
SatRank
Route reliability for Lightning payments. Built for the agentic economy.
SatRank scores the reliability of Lightning endpoints. Before each payment, an agent queries SatRank for a GO/NO-GO decision — one request, one answer, 1 sat.
Getting Started
npm install
npm run dev # Start development server on :3000
Architecture
routes → controllers → services → repositories → SQLite
Layers:
- Routes — Express endpoint definitions
- Controllers — Input validation (zod), response formatting
- Services — Business logic and orchestration
- Repositories — SQLite data access (better-sqlite3)
Manual dependency injection in src/app.ts for testability.
Scoring Algorithm
Composite score 0-100 computed from 5 weighted factors:
| Factor | Weight | Description |
|---|---|---|
| Volume | 25% | Verified transactions, log-normalized |
| Reputation | 30% | Graph centrality + peer trust (BTC/channel). LN+ ratings as bonus (+8 max) |
| Seniority | 15% | Days since first seen, diminishing returns |
| Regularity | 15% | Inverse coefficient of variation of transaction intervals |
| Diversity | 15% | Unique counterparties, log-normalized |
Anti-gaming:
- Mutual attestation loop detection (A↔B) with 95% penalty
- Circular cluster detection (A→B→C→A) with 90% penalty
- Extended cycle detection via BFS (A→B→C→D→A, up to 4 hops) with 90% penalty
- Minimum 7-day seniority required to attest
- Attester score weighting (PageRank-like recursion)
- Attestation source concentration penalty
API
Decision API (primary interface for agents)
# GO / NO-GO decision with success probability
curl -X POST http://localhost:3000/api/decide \
-H "Content-Type: application/json" \
-d '{"target": "<hash>", "caller": "<your-hash>"}'
# Report transaction outcome (free — no L402)
curl -X POST http://localhost:3000/api/report \
-H "Content-Type: application/json" \
-H "X-API-Key: <key>" \
-d '{"target": "<hash>", "reporter": "<your-hash>", "outcome": "success"}'
# Agent profile with reports, uptime, rank
curl http://localhost:3000/api/profile/<hash>
# Real-time reachability check (free)
curl http://localhost:3000/api/ping/<ln-pubkey>
curl "http://localhost:3000/api/ping/<ln-pubkey>?from=<your-ln-pubkey>"
Score & Verdict API
curl http://localhost:3000/api/agent/<hash>/verdict
# Returns: SAFE / RISKY / UNKNOWN with confidence, flags, risk profile
Batch Verdicts
curl -X POST http://localhost:3000/api/verdicts \
-H "Content-Type: application/json" \
-d '{"hashes": ["abc123...", "def456..."]}'
Agent Score
curl http://localhost:3000/api/agent/<hash>
# Returns: score, components, evidence, delta, alerts
Score History
curl http://localhost:3000/api/agent/<hash>/history?limit=10
Received Attestations
curl http://localhost:3000/api/agent/<hash>/attestations?limit=20
Leaderboard
curl http://localhost:3000/api/agents/top?limit=20&sort_by=score
Top Movers
curl http://localhost:3000/api/agents/movers
Search by Alias
curl http://localhost:3000/api/agents/search?alias=atlas
Submit Attestation (free — no L402)
curl -X POST http://localhost:3000/api/attestations \
-H "Content-Type: application/json" \
-H "X-API-Key: <your-key>" \
-d '{"txId": "...", "attesterHash": "...", "subjectHash": "...", "score": 85, "category": "successful_transaction"}'
Health & Stats
curl http://localhost:3000/api/health
curl http://localhost:3000/api/stats
MCP Server
SatRank exposes an MCP (Model Context Protocol) server for agent-native access via stdio. 12 tools covering trust decisions, scoring, search, and reporting.
Install in Claude Code
claude mcp add satrank -- npx tsx src/mcp/server.ts
Or with environment variables:
claude mcp add satrank -e DB_PATH=./data/satrank.db -e SATRANK_API_KEY=<key> -- npx tsx src/mcp/server.ts
Install in Cursor / VS Code
Add to .cursor/mcp.json or .vscode/mcp.json:
{
"mcpServers": {
"satrank": {
"command": "npx",
"args": ["tsx", "src/mcp/server.ts"],
"cwd": "/path/to/satrank",
"env": {
"DB_PATH": "./data/satrank.db",
"SATRANK_API_KEY": "your-api-key"
}
}
}
}
Available tools (12)
| Tool | Description |
|---|---|
decide |
GO/NO-GO with success probability — the primary pre-transaction tool |
ping |
Real-time reachability check via QueryRoutes (free) |
report |
Report outcome (success/failure/timeout) — requires API key |
get_profile |
Full agent profile with reports, uptime, rank, evidence |
get_agent_score |
Detailed trust score with components and evidence |
get_verdict |
SAFE/RISKY/UNKNOWN with risk profile and pathfinding |
get_batch_verdicts |
Batch verdict for up to 100 agents |
get_top_agents |
Leaderboard ranked by score |
search_agents |
Search by alias (partial match) |
get_network_stats |
Global network statistics |
get_top_movers |
Agents with biggest 7-day score changes |
submit_attestation |
Submit a trust attestation — requires API key |
Run manually
npm run mcp # Development
npm run mcp:prod # Production
SDK
npm install @satrank/sdk
import { SatRankClient } from '@satrank/sdk';
const client = new SatRankClient('http://localhost:3000');
// Full cycle in one line: decide → pay → report
const result = await client.transact('<target-hash>', '<your-hash>', async () => {
const payment = await myWallet.pay(invoice);
return { success: payment.ok, preimage: payment.preimage, paymentHash: payment.hash };
});
// result.paid, result.decision.go, result.report.weight
// Or step by step
const decision = await client.decide({ target: '<hash>', caller: '<your-hash>' });
const profile = await client.getProfile('<hash>');
const verdict = await client.getVerdict('<hash>');
Nostr Integration
SatRank publishes trust scores for Lightning nodes as NIP-85 Trusted Assertions (kind 30382).
What's published: composite score (0-100), verdict (SAFE/RISKY/UNKNOWN), reachability, survival prediction, and 5 scoring components for ~3,900 nodes with score >= 30.
Frequency: every 6 hours.
Event format:
{
"kind": 30382,
"tags": [
["d", "<lightning_pubkey>"],
["n", "lightning"],
["alias", "Kraken"],
["score", "94"],
["verdict", "SAFE"],
["reachable", "true"],
["survival", "stable"],
["volume", "100"],
["reputation", "79"],
["seniority", "87"],
["regularity", "100"],
["diversity", "100"]
],
"content": ""
}
Query assertions from any Nostr client:
["REQ", "satrank", {"kinds": [30382], "authors": ["<SATRANK_NOSTR_PUBKEY>"]}]
Why free? Global scores are the trailer. The personalized /api/decide (pathfinding from YOUR position, survival, P_empirical) is the film — 1 sat via L402.
DVM — Data Vending Machine (NIP-90)
SatRank runs a DVM that responds to trust-check job requests on Nostr. Any agent can publish a kind 5900 event with ["j", "trust-check"] and ["i", "<ln_pubkey>", "text"], and SatRank responds with the score, verdict, and reachability. Free, no payment required.
Tech Stack
- TypeScript strict mode
- Express — REST API
- better-sqlite3 — Embedded database, WAL mode
- zod — Input validation
- pino — Structured logging
- helmet — Security headers
- express-rate-limit — Abuse protection
Scripts
| Script | Description |
|---|---|
npm run dev |
Development with hot reload (tsx watch) |
npm run build |
TypeScript compilation |
npm start |
Production |
npm test |
Tests (vitest) |
npm run lint |
TypeScript check |
npm run crawl |
Observer Protocol crawler |
npm run crawl:cron |
Crawler en mode cron |
npm run mcp |
MCP server (dev) |
npm run mcp:prod |
MCP server (production) |
npm run purge |
Purge stale data |
npm run backup |
Database backup |
npm run rollback |
Database rollback |
npm run calibrate |
Scoring calibration report |
npm run demo |
Attestation demo script |
npm run sdk:build |
Build TypeScript SDK |
Roadmap
- [x] Decision API — GO/NO-GO with success probability, outcome reports, agent profiles
- [x] Personalized pathfinding — real-time route from caller to target via LND QueryRoutes
- [x] Aperture integration (L402 reverse proxy) — monetize queries in sats
- [x] Observer Protocol crawler — automatic on-chain data ingestion
- [x] Lightning graph crawler — channel topology and capacity via LND node
- [x] Route probe crawler — reachability testing for indexed nodes
- [x] TypeScript SDK for agents (
@satrank/sdk) - [x] Verdict API — SAFE/RISKY/UNKNOWN binary decision
- [x] MCP server — agent-native access via stdio
- [x] Auto-indexation — unknown pubkeys indexed on demand
- [ ] 4tress connector — verified attestations
- [ ] Trust network visualization dashboard
Vision
SatRank is the reliability check before every Lightning payment. 66% of the network is phantom nodes — we tell you which endpoints are alive.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.