Personal Hermes agent for Adam McBride. Powered by Emblem Vault.

A self-improving agent with persistent memory, multi-chain crypto operations, portfolio tracking, automated workflows, and multi-platform messaging. Built on Nous Research's Hermes agent framework with Emblem Vault integration for 200+ trading tools across 7 blockchains.

• Crypto operations — swap, bridge, and manage tokens across Solana, Ethereum, Base, BSC, Polygon, Hedera, and Bitcoin via Emblem Vault.
• Portfolio tracking — automated snapshots, P&L tracking, allocation drift alerts, and daily performance summaries.
• Persistent memory — SQLite-backed memory with full-text search. Remembers preferences, facts, and conversation context across sessions.
• Scheduled automations — morning briefings, portfolio checks every 4 hours, news digests, and memory nudges run unattended.
• Multi-platform — CLI, Telegram, Discord. Start on one, pick up on another.
• Research — token lookup, market overview, safety checks, and news aggregation.
• Task management — notes, to-dos, reminders, and daily briefings.
• Self-improving — auto-memorizes important interactions, builds behavioral patterns over time.

git clone https://github.com/Adam-McBride/Emblem-AI.git
cd Emblem-AI
pip install -e .
cp .env.example .env
# Fill in your API keys in .env
python emblem_ai.py

Copy .env.example to .env and set:

• OPENROUTER_API_KEY — LLM provider key (OpenRouter, OpenAI, or Anthropic).
• EMBLEM_API_KEY — Emblem Vault API key for crypto operations.
• EMBLEM_WALLET_PASSWORD — wallet password for transaction signing.
• TELEGRAM_BOT_TOKEN — optional, for Telegram gateway.
• DISCORD_BOT_TOKEN — optional, for Discord gateway.

emblem_ai.py (entry point)
├── config/
│   ├── agent_config.py  — environment and config loading
│   └── persona.py       — personality and behavioral traits
├── skills/
│   ├── crypto/
│   │   ├── wallet.py    — multi-chain wallet via Emblem Vault
│   │   ├── trading.py   — swaps, limit orders, trade management
│   │   └── bridge.py    — cross-chain bridging
│   ├── portfolio/
│   │   └── tracker.py   — snapshots, P&L, allocation alerts
│   ├── research/
│   │   └── market.py    — token lookup, news, safety checks
│   ├── assistant/
│   │   └── tasks.py     — notes, to-dos, daily briefings
│   └── scheduling/
│       └── scheduler.py — cron automations
├── memory/
│   └── store.py         — SQLite + FTS5 persistent memory
├── gateway/
│   └── adapters.py      — CLI, Telegram, Discord adapters
├── workflows/
│   └── automations.py   — multi-step automated workflows
└── tools/
    └── llm_client.py    — multi-provider LLM client

The agent uses a defined persona with:

• Behavioral traits (direct, proactive, cautious with money, crypto-native).
• Boundaries (never share keys, always confirm large trades, no financial advice).
• Proactive behaviors (alert on portfolio changes, flag gas spikes, summarize overnight activity).

• X: @EmblemAI_
• Builder: @adamamcbride
• Powered by: Emblem Vault

MIT