A GoBot on Solana — autonomous trading intelligence with a virtual pet soul.
Physical companion: TamaGOchi (Arduino Modulino® I2C) · By NanoSolana Labs
Point the docs page at a running NanoBot instance and it will stream the current daemon, wallet, OODA, and recent trade snapshot directly from the local API.
{}
The docs now match the current daemon: local-first MiniMax, native Claude support, cleaner Telegram ops, remote gateway/Tailscale access, and an upgraded live status view.
minimax-m2.7:cloud is now the default local Ollama harness model for normal chat,
/deepsolana, and skill-aware Solana assistance.
Anthropic is wired in as a native backend, and Claude Code session workflows still remain available for repo creation, coding, and GitHub automation.
Inline menus, callback actions, smarter natural language, GitHub creation fixes, /personality,
and a cleaner /status are now part of the default bot UX.
The gateway now supports Tailscale Serve/Funnel patterns and remote-over-SSH setup modeled on the OpenClaw-style control path.
/web and /xsearch now route through the proper Grok tool model instead of the
broken generic chat payload path.
/status now reports daemon runtime, active model, wallet, Honcho, Aster, Hyperliquid, and
service health without exposing raw Solana program failure text.
From market data ingestion to hardware LED feedback — one binary does it all.
Observe → Orient → Decide → Act. Continuous autonomous cycle pulling Helius RPC, Birdeye OHLCV, and Aster perps data.
3-tier epistemological memory: Known (fresh API data) → Learned (trade patterns) → Inferred (cross-asset correlations).
Telegram sessions become structured trading memory: summaries, peer cards, conclusions, and recalled user preferences for Solana spot and perps.
Multi-condition signal engine with auto-optimizer. Parameters adjust based on rolling trade performance.
Monetize your bot's API with crypto-gated endpoints. Multi-chain USDC payments via Solana, Base, Polygon.
8× RGB LEDs, buzzer, 3 buttons, rotary knob, IMU, thermometer, ToF sensor. Physical trading feedback.
Native ARM64 binary for Orin Nano and Spark. Deploy via Brev.dev GPU cloud or bare-metal JetPack.
Auto-generates and persists a Solana keypair on first boot. Standard keygen format, 0600 permissions.
Telegram Bot API integration with long polling, inline dashboards, operator personalities, strict
TELEGRAM_ID access control, and natural-language Aster, GitHub, and research workflows.
Virtual pet evolves based on on-chain performance: Egg → Larva → Juvenile → Adult → Alpha. Mood driven by P&L.
EVM perpetuals on Hyperliquid. Open, close, and cancel positions from Telegram with EIP-712 L1 action signing — no web UI needed.
A single daemon orchestrates wallet, RPC, pet engine, Telegram, x402, and hardware — all in <10MB.
Honcho turns Telegram chat turns into durable Solana trading memory. The daemon now creates workspace-aware sessions, models peers, writes trading conclusions, and recalls the right summary before the next trade or research step.
User messages, bot replies, wallet context, watchlists, and session mode are stored in a Honcho session.
Honcho builds summaries, peer cards, and representations for the user and the agent inside the trading workspace.
Repeated preferences become explicit conclusions: preferred tokens, spot vs perps bias, and risk posture.
Before responding, the bot injects summaries and matched conclusions back into prompt context for the next decision.
Loading memory turn…
Waiting for summary…
No conclusion yet.
Recalled context will appear here.
# Inspect the new memory layer /honcho_status → workspace, model, and bridge status /honcho_context → messages + summaries + recalled conclusions /honcho_sessions → list tracked trading sessions /honcho_summaries → short and long session summaries /honcho_search bonk scalps → search messages inside the active session /honcho_messages → recent stored turns /honcho_conclusions BONK → semantic trading conclusions
Software intelligence bridged with physical hardware. 7 sensors over a single Qwiic I2C cable.
8× RGB LEDs — LC8822
Status: idle 🔵 signal 🟣 trade 🟡 win 🟢 loss 🔴PKLCS1212E — Tone Generator
Audio alerts for signals, trades, wins, errors3× Push + LED
A = trigger cycle · B = toggle mode · C = e-stopPEC11J — Rotary Encoder
Real-time RSI threshold tuningLSM6DSOX — 6-axis IMU
Tilt detection → auto-pause tradingHS3003 — Temp + Humidity
Environment logging to ClawVaultVL53L4CD — Time-of-Flight
Proximity wake-up (<5cm triggers cycle)make orin
brev create nanosolana --gpu
make rpi
make macos
Your GoBot has a soul. A virtual pet whose life is driven by real on-chain performance. Physical companion: TamaGOchi (Arduino Modulino® I2C).
🥚 NanoSolana 😐 📊 Stage: egg · Level 1 · XP 0 😐 Mood: neutral ⚡ Energy: ⚡⚡⚡⚡⚡⚡⚡⚡⚡⚡ 🍽️ Hunger: 🟢🟢🟢🟢🟢🟢🟢🟢🟢🟢 📈 Trades: 0 · Win Rate: 0% 💰 Balance: 0.0000 SOL 📊 Total PnL: +0.0000 SOL 🔥 Streak: +0 ⏱️ Age: 0s · Uptime: 0h
Mood is driven by streak, PnL, and wallet balance in real-time.
Monetize your GoBot's intelligence. Accept USDC payments per API call across multiple chains.
GET /health — free GET /x402/info — free GET /api/signals — 0.001 USDC per call GET /api/research — 0.001 USDC per call GET /api/agent — 0.001 USDC per call
Full Hyperliquid perpetuals integration — open longs and shorts, manage positions, and cancel orders directly from Telegram. EIP-712 L1 action signing with msgpack encoding, no web UI required.
# Setup in .env HYPERLIQUID_PRIVATE_KEY=0x... HYPERLIQUID_WALLET=0x... # Account overview /hl → account value, margin, open positions /hl_balance → quick balance + withdrawable /hl_positions → open perps: side, size, entry, PnL, leverage /hl_orders → open limit orders # Trading /hl_open BTC long 0.01 → open BTC long, 0.01 BTC, 5% slippage /hl_open ETH short 0.1 2% → open ETH short with 2% slippage /hl_close BTC → close full BTC position /hl_close SOL 10 → partial close 10 SOL /hl_cancel → cancel all open orders
The live Go runtime now defaults to minimax-m2.7:cloud via Ollama. It sits inside the normal
NanoSolana harness, so system prompt, Solana context, Honcho memory, and skill injections all reach the same
local-first model path.
# Pull and run the default local harness model ollama run minimax-m2.7:cloud # Or pull first, run later ollama pull minimax-m2.7:cloud
# Claude Code ollama launch claude --model minimax-m2.7:cloud # Codex ollama launch codex --model minimax-m2.7:cloud # OpenCode ollama launch opencode --model minimax-m2.7:cloud # OpenClaw ollama launch openclaw --model minimax-m2.7:cloud
# Make MiniMax the daemon default LLM_PROVIDER=ollama OLLAMA_MODEL=minimax-m2.7:cloud OLLAMA_BASE_URL=http://127.0.0.1:11434 # Optional native Claude backend ANTHROPIC_API_KEY=sk-ant-... ANTHROPIC_MODEL=claude-sonnet-4-6 # Restart daemon to pick up nanosolana daemon
| Layer | Current behavior | Why it matters |
|---|---|---|
| Default routing | Normal chat resolves to Ollama first | MiniMax is the live default |
| System prompt | Uses the existing NanoSolana harness prompt | Solana tools and skills still apply |
/deepsolana |
Targets the configured Ollama harness model | No hardcoded model branch |
/model |
Accepts bare Ollama refs like minimax-m2.7:cloud |
Cleaner operator switching |
| Hosted providers | Anthropic and xAI remain available | Local-first without losing fallbacks |
Install NanoSolana from GitHub or the published npm package, then run nanosolana from anywhere.
curl -fsSL https://raw.githubusercontent.com/x402agent/nanosolana-go/main/install.sh | bash
curl -fsSL https://raw.githubusercontent.com/x402agent/nanosolana-go/main/install.sh | bash -s -- --with-web
npx nanosolana-cli install --with-web
npm install -g nanosolana-cli nanosolana version
Installs: ~/nanosolana/build/nanosolana, a stable CLI at
~/.nanosolana/bin/nanosolana, the agent wallet, ClawVault workspace, and the optional web console.
$ git clone https://github.com/x402agent/nanosolana-go.git $ cd nanosolana-go $ cp .env.example .env $ make build $ ./build/nanosolana version
# Edit .env with your API keys HELIUS_API_KEY=your_key_here HELIUS_RPC_URL=https://mainnet.helius-rpc.com/?api-key=KEY BIRDEYE_API_KEY=optional
$ nanosolana daemon 🐹 TamaGOchi — A GoBot on Solana 🔑 Agent wallet: 7xKXqR8...3vBp 🤖 Telegram: @YourBot connected 💰 x402 paywall: :18402
Package your skill folder and publish it to hub.nanosolana.com via npm.
npx @nanosolana/nanohub login
npx @nanosolana/nanohub publish ./my-skill \ --slug my-skill \ --name "My Skill" \ --version 1.0.0 \ --tags latest,solana
Requirements: SKILL.md (or skills.md), text files only, and a
semver version.