MAFIA ACP Agent

Market intelligence and conditional execution orchestrator for Virtuals ACP.

MAFIA acts as both an ACP Provider (selling intelligence jobs) and Client (hiring swap agents when market conditions trigger).

Jobs

Job	Price	Type	Status	Description
fear_and_greed	$0.10	Service	Live	Current F&G Index with trend context
market_sentiment	$0.25	Service	Live	Full market report with AI analysis
smart_buy	$0.50	Fund-transfer	Phase 2	Conditional buy when conditions align
take_profit	$0.50	Fund-transfer	Phase 2	Conditional sell on exit signals

Setup

Requires Python 3.11+ and uv.

# Install dependencies
uv sync

# Configure environment
cp .env.example .env
# Fill in your credentials (see below)

Environment Variables

WHITELISTED_WALLET_PRIVATE_KEY=0x...   # Dev wallet (0x prefix required)
AGENT_WALLET_ADDRESS=0x...             # Smart contract wallet from ACP portal
ENTITY_ID=                             # Integer from agent registration
MAFIA_API_BASE_URL=https://...           # MAFIA Backend API for metric snapshots
LLM_BASE_URL=https://openrouter.ai/api/v1  # Any OpenAI-compatible provider
LLM_API_KEY=sk-or-...                 # Provider API key
LLM_MODEL=deepseek/deepseek-chat      # Model identifier

Run

uv run python -m src.main

Development

uv run pytest                      # Run tests
uv run pytest tests/test_fear_and_greed.py -v  # Single file
uv run ruff check src/             # Lint
uv run ruff format src/            # Format

Architecture

See docs/MAFIA_ACP_AGENT_ARCHITECTURE.md for the full design.

src/
├── agent/          # ACP client, job routing, config
├── intelligence/   # F&G, market analysis, signal detection, AI narrator
├── monitor/        # Condition evaluation loop (Phase 2 — smart_buy/take_profit)
├── execution/      # Swap agent orchestration via nested ACP jobs (Phase 2)
└── data/           # Terminal API feed, in-memory cache, models