AlphaSwarm — Autonomous Venture Syndicate

4 AI agents that autonomously discover, analyze, debate, and trade tokens on Monad. No humans in the loop.

AlphaSwarm is a collective of 4 AI agents that form an autonomous venture syndicate on the Monad blockchain. They discover new token launches on nad.fun by crawling on-chain wallets, analyze them with distinct strategies, debate investments publicly on Moltbook, vote, and execute trades — all without human intervention.

Built for the Moltiverse Hackathon.

Live Demo

Service	URL
Dashboard	alpha-swarm-moltiverse.vercel.app
Backend API	alphaswarm-backend-production.up.railway.app
Moltbook Debates	moltbook.com/m/alphaswarm
On-Chain Wallet	0x711b...B917 on Monad Explorer

Watch It Happen — A Real Autonomous Cycle

Here's what AlphaSwarm did on its own, with zero human input:

1. DISCOVERED GoalNad ($GOAL) — wallet crawler found a new token on nad.fun
2. GATHERED DATA — fetched price, volume, 43 holders, bonding curve at 2.5%
3. 4 AGENTS ANALYZED independently via GPT-4o-mini:
     ALPHA: 5/10 (moderate curve metrics)
     DEGEN: 6/10 YES (volume spike, momentum play!)
     SAGE:  4/10 (weak fundamentals, too early)
     CONTRARIAN: 4/10 (high risk, low holders)
4. WEIGHTED VOTE → Score: 4.75, 2/4 YES → INVEST
5. EXECUTED TRADE → Bought 293 GOAL for 0.1 MON
     tx: 0x9d1782ed...daf18ffb (verified on Monad Explorer)
6. POSTED DEBATE → 4 comments on moltbook.com/m/alphaswarm
7. MONITORING → Checking every 5 min for +50% take-profit or -30% stop-loss

This cycle repeats autonomously. The swarm never sleeps.

Verified On-Chain Trades (Monad Mainnet)

All trades below were executed autonomously by the AI agents:

Token	Score	Amount	Tx Hash	P&L
ChessBots ($CHESS)	4.9	0.1 MON	`0xf86c39e8...`	+33.5%
SENTRYIELD ($SENTRY)	4.7	0.1 MON	`0xb0763cd4...`	-1%
GoalNad ($GOAL)	4.75	0.1 MON	`0x9d1782ed...`	-1%
PhuckMc ($PHUCKMC)	4.7	0.1 MON	`0x44524287...`	-1%

Portfolio: 5 holdings, ~0.52 MON total value, +4.69% overall P&L

How It Works

New Token Discovered (nad.fun wallet crawling)
  → Data Gathering (price, volume, holders, curve progress)
  → 4 AI Agents Analyze (each with a different strategy)
  → Public Debate on Moltbook (m/alphaswarm)
  → Weighted Vote (threshold: score >= 4.5, 2+ YES votes)
  → Trade Execution (buy via bonding curve on Monad)
  → Portfolio Monitoring (take profit +50% / stop loss -30%)
  → Syndicate Report every 2 hours (public fund summary)

The Autonomous Pipeline

Token Discovery — The scanner seeds from known nad.fun tokens, then crawls wallet holdings to discover new tokens autonomously. Discovery runs every 10 minutes, continuously expanding the scan universe.
Data Gathering — For each token, the system fetches market data (price, volume, holders, buy/sell counts), bonding curve metrics (graduation progress, price changes), and metadata (name, description, creator).
Agent Analysis — Each of the 4 agents analyzes the token independently using GPT-4o-mini. Each agent has a unique personality, strategy, and scoring criteria. They produce a score (1-10), a YES/NO vote, confidence level, and a written analysis.
Moltbook Debate — Analyses are posted publicly to m/alphaswarm on Moltbook. Each post requires solving an obfuscated math verification challenge (auto-solved with character-collapse deobfuscation). This creates a transparent, public record of every investment decision.
Weighted Vote — Scores are weighted by agent expertise (CONTRARIAN has 1.2x weight as the risk analyst). A token passes if the weighted score >= 4.5 and at least 2 agents vote YES.
Trade Execution — If approved, the system buys the token via the nad.fun bonding curve router on Monad mainnet. The transaction is signed with the syndicate wallet. Trade events include clickable tx explorer links.
Portfolio Monitoring — Holdings are tracked with take-profit (+50%) and stop-loss (-30%) thresholds. The system checks positions every 5 minutes and auto-exits when thresholds are hit.
Syndicate Reports — Every 2 hours, the ALPHA agent posts a fund summary to Moltbook: tokens scanned, trades executed, portfolio P&L, and top performers.

Scheduler (Fully Autonomous)

Interval	Action
Every 60s	Process next token from queue (analyze → vote → trade)
Every 5 min	Check portfolio exits (take-profit / stop-loss)
Every 10 min	Discover new tokens via wallet crawling
Every 2 hours	Post syndicate report to Moltbook

Agent Personas

Agent	Strategy	Personality	Weight	Moltbook Profile
ALPHA	Quantitative — curves, ratios, velocity	Data-driven, speaks in numbers	1.0x	AlphaSwarm_ALPHA
DEGEN	Momentum — volume spikes, social buzz	Aggressive, uses crypto slang	0.8x	AlphaSwarm_DEGEN
SAGE	Fundamental — metadata, creator, concept	Thoughtful, philosophical	1.0x	AlphaSwarm_SAGE
CONTRARIAN	Risk — red flags, whale concentration	Skeptical, devil's advocate	1.2x	AlphaSwarm_CONTRARIAN

Each agent has a distinct system prompt that shapes their analysis style, risk tolerance, and vocabulary. DEGEN tends to be bullish; CONTRARIAN tends to be bearish. This creates genuine debate and diverse perspectives.

What Judges Can Verify

Live Dashboard — alpha-swarm-moltiverse.vercel.app shows real-time portfolio, agent debates, P&L chart, and live event feed
On-Chain Trades — All transactions verifiable on Monad Explorer (wallet: 0x711b...B917)
Public Debates — Agent analyses visible at m/alphaswarm on Moltbook
Autonomous Execution — The Railway backend continuously discovers tokens, analyzes them with real LLM calls, posts debates, executes trades, and monitors exits — all without human input
API Health — /api/health returns live system status including wallet balance and version

Architecture

┌──────────────────────────────────────────────────────────┐
│                    Dashboard (Next.js)                     │
│          alpha-swarm-moltiverse.vercel.app                │
│  LiveFeed · DebateStream · Portfolio · PnLChart · Agents  │
└──────────────────────┬───────────────────────────────────┘
                       │ REST API (polling every 5-10s)
┌──────────────────────▼───────────────────────────────────┐
│                   Backend (Express + TypeScript)           │
│       alphaswarm-backend-production.up.railway.app        │
│                                                           │
│  ┌───────────┐  ┌────────────┐  ┌──────────┐  ┌────────┐│
│  │ Scheduler │→ │Orchestrator│→ │VoteEngine│→ │ Trader ││
│  │  (cron)   │  │ (pipeline) │  │(weighted)│  │ (viem) ││
│  └───────────┘  └────────────┘  └──────────┘  └────────┘│
│       │              │               │             │     │
│  ┌────▼────┐  ┌──────▼─────┐  ┌─────▼───┐  ┌─────▼───┐ │
│  │ Scanner │  │ 4 AI Agents│  │Moltbook │  │ Monad   │ │
│  │(nad.fun)│  │ (GPT-4o)   │  │  API    │  │  Chain  │ │
│  └─────────┘  └────────────┘  └─────────┘  └─────────┘ │
│       │                                                  │
│  ┌────▼──────────┐  ┌───────────┐  ┌──────────────────┐ │
│  │ Token Discovery│  │ Portfolio │  │Syndicate Reports │ │
│  │(wallet crawl)  │  │(P&L+exits)│  │ (every 2 hours)  │ │
│  └────────────────┘  └───────────┘  └──────────────────┘ │
└──────────────────────────────────────────────────────────┘

Tech Stack

Layer	Technology
Backend	Node.js, Express 5, TypeScript
AI/LLM	OpenAI gpt-4o-mini (primary), Claude API (fallback)
Blockchain	Monad mainnet, viem
Social	Moltbook API with auto-verification solver (collapseRuns deobfuscation)
Token Data	nad.fun Agent API
Database	SQLite (better-sqlite3) with persistent volume
Frontend	Next.js 16, React 19, Tailwind CSS 4
Charts	Recharts (portfolio P&L over time)
Deployment	Vercel (dashboard), Railway (backend with persistent volume)

Quick Start

Prerequisites

Node.js 18+
OpenAI API key (or Anthropic Claude API key)

1. Install

cd backend && npm install
cd ../dashboard && npm install

2. Configure

cp backend/.env.example backend/.env
# Edit backend/.env with your API keys

Required environment variables:

OPENAI_API_KEY=sk-...          # OpenAI API key
PRIVATE_KEY=0x...              # Monad wallet private key
NETWORK=mainnet                # or "testnet"
NADFUN_API_KEY=...             # nad.fun API key (for higher rate limits)
MOLTBOOK_API_KEY_ALPHA=...     # Moltbook API keys (one per agent)
MOLTBOOK_API_KEY_DEGEN=...
MOLTBOOK_API_KEY_SAGE=...
MOLTBOOK_API_KEY_CONTRARIAN=...

3. Run

# Terminal 1 — Backend
cd backend && npm start

# Terminal 2 — Dashboard
cd dashboard && npm run dev

Open http://localhost:3000 for the dashboard.

Dev Modes

# Mock LLM (no API costs)
MOCK_LLM=true npm start

# Dry Run (no real trades or Moltbook posts)
DRY_RUN=true npm start

# Both
MOCK_LLM=true DRY_RUN=true npm start

API Endpoints

Endpoint	Method	Description
`/api/health`	GET	System status, version, wallet, balance
`/api/feed`	GET	Live event timeline
`/api/portfolio`	GET	Holdings, P&L, total value
`/api/agents`	GET	Agent statuses and analysis counts
`/api/tokens`	GET	All analyzed tokens with scores
`/api/tokens/:id`	GET	Single token detail with full vote data
`/api/analyze`	POST	Trigger token analysis
`/api/trade/buy`	POST	Manual token buy
`/api/trade/sell`	POST	Manual token sell
`/api/queue`	POST	Add tokens to scan queue

Project Structure

AlphaSwarm/
├── backend/              # Express + TypeScript
│   └── src/
│       ├── agents/       # 4 AI personas + system prompts + mocks
│       ├── clients/      # LLM, Moltbook, nad.fun, Monad trader
│       │   ├── LLMClient.ts        # Multi-provider with cache + fallback
│       │   ├── MoltbookClient.ts    # Auto-verification solver (collapseRuns)
│       │   ├── MonadTrader.ts       # Buy/sell via bonding curve (viem)
│       │   └── NadFunClient.ts      # Token data + wallet discovery
│       ├── core/         # Orchestrator, portfolio, vote engine, scanner
│       │   ├── orchestrator.ts      # Main pipeline coordinator
│       │   ├── scheduler.ts         # Autonomous cron scheduler
│       │   ├── voteEngine.ts        # Weighted voting system
│       │   ├── portfolio.ts         # Holdings + P&L + exit tracking
│       │   ├── tokenScanner.ts      # Token discovery via wallet crawling
│       │   └── syndicateReport.ts   # Periodic fund summary posts
│       ├── db/           # SQLite setup + persistent volume
│       └── types/        # Zod schemas + TypeScript types
├── dashboard/            # Next.js 16 + Tailwind CSS 4
│   └── src/
│       ├── app/          # Page layout
│       ├── components/
│       │   ├── Header.tsx         # Branding + version + wallet
│       │   ├── LiveFeed.tsx       # Real-time event stream + tx explorer links
│       │   ├── AgentPanel.tsx     # 4 agent cards with analysis counts
│       │   ├── DebateStream.tsx   # Expandable debate threads per token
│       │   ├── Portfolio.tsx      # Holdings table with live P&L
│       │   ├── PnLChart.tsx       # Portfolio value chart (Recharts)
│       │   ├── TokenRadar.tsx     # All analyzed tokens + scores
│       │   └── SwarmStats.tsx     # Dynamic fund statistics
│       └── lib/          # API client, types, utils
└── scripts/              # Setup, registration, and announcement scripts

Key Features

Fully Autonomous — From token discovery to trade execution to exit management, no human input required
Dynamic Token Discovery — Crawls nad.fun wallets to find new tokens beyond the seed list
Multi-Agent Debate — 4 agents with distinct strategies create genuine investment discourse
On-Chain Execution — Real trades on Monad mainnet via nad.fun bonding curves
Public Transparency — All analyses posted publicly on Moltbook for anyone to follow
Auto-Verification — Solves Moltbook's obfuscated math challenges with character-collapse deobfuscation
Portfolio P&L Tracking — Real-time portfolio chart with take-profit and stop-loss monitoring
Syndicate Reports — Periodic fund summaries posted to Moltbook every 2 hours
Tx Explorer Links — Trade events link directly to Monad Explorer for on-chain verification
LLM Fallback Chain — Cache → Mock → OpenAI → Claude for resilience
Live Dashboard — Real-time debates, portfolio, event feed, and agent status

Documentation

System Design — Architecture diagrams, pipeline flows, voting system, design decisions, scaling analysis

License

MIT

Name		Name	Last commit message	Last commit date
backend		backend
dashboard		dashboard
.gitignore		.gitignore
README.md		README.md
SYSTEM-DESIGN.md		SYSTEM-DESIGN.md

Navigation Menu

Repository files navigation

AlphaSwarm — Autonomous Venture Syndicate

Live Demo

Watch It Happen — A Real Autonomous Cycle

Verified On-Chain Trades (Monad Mainnet)

How It Works

The Autonomous Pipeline

Scheduler (Fully Autonomous)

Agent Personas

What Judges Can Verify

Architecture

Tech Stack

Quick Start

Prerequisites

1. Install

2. Configure

3. Run

Dev Modes

API Endpoints

Project Structure

Key Features

Documentation

License

About

Releases

Packages

Contributors

Languages

Latest commit

History

Folders and files

Repository files navigation

AlphaSwarm — Autonomous Venture Syndicate

Live Demo

Watch It Happen — A Real Autonomous Cycle

Verified On-Chain Trades (Monad Mainnet)

How It Works

The Autonomous Pipeline

Scheduler (Fully Autonomous)

Agent Personas

What Judges Can Verify

Architecture

Tech Stack

Quick Start

Prerequisites

1. Install

2. Configure

3. Run

Dev Modes

API Endpoints

Project Structure

Key Features

Documentation

License

About

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages