clawdbot-go
Advanced tools
Zero Clawd — one-shot install for sovereign Solana + Robinhood omni agent runtime: RH skill pack (23), agents, packages, clawdbot CLI hooks. Connect at cheshireterminal.ai/zeroclawd
Product · Agent hub · npm · Security · Release checklist
Zero Clawd is a small, self-hostable runtime for AI agents that need to work on Solana — trading loops, wallets, market data, skills, and a local web console — without a heavy cloud stack.
It is Solana-first. Robinhood Chain / EVM support is additive through an open skill pack (token launch, Uniswap, registries, payments). You bring your own keys and risk limits; live money paths stay opt-in.
| Piece | What it is |
|---|---|
CLI (clawdbot) | Go binary for agent chat, OODA loops, catalog, doctor, status, web console |
npm package (clawdbot-go) | One-shot install + RH skill pack into ~/.clawdbot (and common agent skill dirs) |
| Skill pack | 23 open skills under skills/ — FunPump launch, swaps, LP, Cheshire registries, zk-omni |
| Web console | Local dashboard/API (default 127.0.0.1:18800) you can also wire to cheshireterminal.ai/zeroclawd |
| Laws + DNA | Six-law harness, onchain-oriented identity, and starter agent DNA at install |
| You see… | Means… |
|---|---|
| Zero Clawd | Product name (docs, DNA, UI, install banners) |
clawdbot / clawdbot-go | CLI binary and npm package (kept for compatibility) |
| Zero-Bruh | This GitHub repository |
cheshire-terminal-agents | Separate npm package: agent catalog + dual-chain forge SDK (not the Go runtime) |
make package, stripped CLI on the order of ~10 MBJump to: One-shot install · Quick Start · Architecture · Six laws · CLI · RH skill pack
Published package: clawdbot-go on npm
(npm i clawdbot-go / npm i -g clawdbot-go).
Skills are prepackaged at install — every skills/*/SKILL.md in this repo
(pack-index.json v2, 23 skills) is copied to ~/.clawdbot/skills and
symlinked into ~/.agents, ~/.claude, and ~/.codex skill roots.
npm i clawdbot-go (recommended)# Local project dependency (postinstall prepackages the skill pack)
npm i clawdbot-go
# Global CLI bins: clawdbot-go · zero-clawd · clawdbot-stack
npm i -g clawdbot-go
# If your npm blocks install scripts (skills will not prepackage until allowed):
npm install -g --allow-scripts=clawdbot-go clawdbot-go
# or once for all installs of this package:
# npm config set allow-scripts=clawdbot-go --location=user
Package page: https://www.npmjs.com/package/clawdbot-go
# Explicit skills-only re-prepackage
npx clawdbot-go skills-install --force
# → ~/.clawdbot/skills (+ agent root symlinks)
# From this checkout instead of npm:
cd /path/to/go-bot && npm install
Skip skill prepackage: CLAWDBOT_SKIP_SKILLS=1 npm i clawdbot-go
curl -fsSL https://raw.githubusercontent.com/Solizardking/Zero-Bruh/main/install-npm.sh | bash
# same as:
npx clawdbot-go install
# full stack inside postinstall:
CLAWDBOT_ONESHOT=1 npm i -g clawdbot-go
Also seeds the RH skill pack into ~/.clawdbot/skills and agent skill roots:
curl -fsSL https://raw.githubusercontent.com/Solizardking/Zero-Bruh/main/install.sh | bash
# or from this checkout:
./install.sh
# Skill Hub pack (same Cheshire/RH suite as catalog skills)
npx github:Solizardking/skills install cheshire-terminal-agents --force
# Agents SDK + forge + registries (npm) — catalog/identity, NOT this Go runtime
npm i cheshire-terminal-agents
cheshire-terminal-agents
(v1.48+) ships an optional bridge to this package — it does not hard-depend
on clawdbot-go and does not run Zero Clawd on its postinstall:
# From the agents package: discover + install Zero Clawd (invokes npx clawdbot-go …)
npx cheshire-terminal-agents clawdbot-info
npx cheshire-terminal-agents clawdbot-install --dry-run
npx cheshire-terminal-agents clawdbot-install # → npx clawdbot-go install
npx cheshire-terminal-agents clawdbot-install --skills-only
npx cheshire-terminal-agents clawdbot-install --global # → npm i -g clawdbot-go
# External catalog entry (discoverable, not vendored under agents packages/)
npx cheshire-terminal-agents packages-list
# → packages: clawd-agent-tui, headless-agent, …
# → external: ["clawdbot-go"] (npmUrl → https://www.npmjs.com/package/clawdbot-go)
| Package | npm | Role |
|---|---|---|
| This runtime | clawdbot-go | Skills oneshot, CLI bins, web console :18800 |
| Agents / forge | cheshire-terminal-agents | Catalog, dual-chain forge, nested TS packages + optional clawdbot-install bridge |
Install both when you need identity forge and a local Zero Clawd console.
/zeroclawd)Zero Clawd (this repo) is the local runtime. Cheshire Terminal hosts the public install + connect hub that talks to your agent after the one-shots.
| Surface | Where |
|---|---|
| Product hub | cheshireterminal.ai/zeroclawd |
| Aliases | /clawdbot-go · /clawdbot · /zero-clawd |
| Agent hub / forge | /agents · /agents/forge |
| CT client source | cheshire-terminal/client/src — pages/ClawdbotGoPage.tsx, lib/zeroClawd.ts, routes in App.tsx |
| Install scripts | install-npm.sh · install.sh (printed steps end with the CT connect loop) |
| Local console | clawdbot web → http://127.0.0.1:18800 (web/backend + web/frontend) |
# One-shots (also the curl lines on the hosted hub)
curl -fsSL https://raw.githubusercontent.com/Solizardking/Zero-Bruh/main/install-npm.sh | bash
curl -fsSL https://raw.githubusercontent.com/Solizardking/Zero-Bruh/main/install.sh | bash
# from this checkout:
./install-npm.sh
./install.sh
# Start agent API/console + allow browser-direct probes from Cheshire
export CLAWDBOT_CORS_ORIGINS=https://cheshireterminal.ai
clawdbot web # or: go run ./web/backend -port 18800
# → open https://cheshireterminal.ai/zeroclawd and Connect http://127.0.0.1:18800
Hosted probes (browser-direct for loopback; /api/zeroclawd/probe for public agents):
| Endpoint | Role |
|---|---|
GET /api/health | Liveness · agent Zero Clawd · package clawdbot-go · product URL |
GET /api/status | Runtime status (+ public_links ecosystem map) |
GET /api/dna | Starter DNA |
GET /api/ecosystem | Product/repo surfaces |
GET /api/rh/readiness | RH 4663 presence gate |
Also: live npm meta via Cheshire GET /api/zeroclawd/npm → registry.npmjs.org/clawdbot-go/latest.
Loopback agents are probed from your browser (not the Cheshire server). Public remote hosts use the same-origin probe bridge (SSRF-hardened — private/link-local/loopback blocked server-side).
| What you get | Path |
|---|---|
| RH skill pack (23) | ~/.clawdbot/skills + ~/.agents/skills/* (prepackaged) |
| Pack manifest | ~/.clawdbot/skills/.clawdbot-prepackaged.json |
| Env template | ~/.clawdbot/.env (CLAWDBOT_SKILLS_DIR, RH hooks) |
| npm package | clawdbot-go · npm i clawdbot-go |
| Agents bridge | cheshire-terminal-agents · npx cheshire-terminal-agents clawdbot-install (optional; not a hard dep) |
| CLI binaries | clawdbot-go / zero-clawd (npm) · clawdbot (Go, via install.sh) |
| Hosted connect | cheshireterminal.ai/zeroclawd |
| Product | zeroclawd · agents · forge · funpump.ai |
export CLAWDBOT_SKILLS_DIR="$HOME/.clawdbot/skills"
npx clawdbot-go skills # list pack ids
npx clawdbot-go inspect # validate SKILL.md files
npx clawdbot-go skills-install # re-prepackage all skills
clawdbot catalog skills # if Go binary present
This tree is a working, verifiable trading agent — not just a dashboard.
Solana runtime stays first-class; Robinhood Chain / EVM is additive via the
bundled skill pack and pkg/rh.
Robinhood Crypto Agent Open Stack (v2 · 23 skills). Vendored under
skills/ for FunPump launch, Uniswap, strategy bots, payments,
Cheshire ERC-8004 registries, omni mint, zk-omni, and Blockscout web3-dev.
Product host: funpump.ai · Zero Clawd:
cheshireterminal.ai/zeroclawd · agent hub:
cheshireterminal.ai/agents · forge:
cheshireterminal.ai/agents/forge.
Open agents package: npm ·
GitHub ·
SkillHub.
Core operator env: BLOCKSCOUT_API_KEY + RH_RPC_URL (chain 4663).
See docs/RH_CRYPTO_AGENT_STACK.md and
skills/README.md.
RH readiness in the Go runtime. pkg/config loads RH settings;
pkg/rh builds JSON-RPC + Blockscout PRO requests; clawdbot doctor reports
connectors.robinhood; web console exposes /api/connectors and
GET /api/rh/readiness (presence only — never secret values).
Live market data. Web console pulls Jupiter prices (/api/market/prices),
Birdeye perps OI and trending when keys allow, plus strategy / backtest panels.
Trading engine. Risk-based position sizing, portfolio limits (drawdown
circuit breaker), and a backtest harness that replays the same Evaluate() as
the live OODA loop.
This codebase is a forked descendant of three foundational repositories that defined the academic lineage of compression, encryption, cellular automata, multi-agent systems, and algorithmic game theory.
The docs/PiedPiper-master/ directory is a verbatim archive of
vs666/MinMax, a landmark project that
implemented data compression (Huffman, Arithmetic, BWT+RLE),
encryption (AES-128, DES, RSA, cellular-automaton-based PRNG),
Conway's Game of Life, multi-agent collision avoidance, and
cryptographic hash optimization from first principles.
Clawd inherits three direct code descendants:
| PiedPiper Source | Clawd Package | Description |
|---|---|---|
GameOfLife/ | pkg/gameoflife/ | Toroidal Life engine — the universal computer |
Compression/ (middle-out) | pkg/middleout/ | Content-cache, Ralph loop, content router |
Compression/ (Weissman score) | pkg/weissman/ | Compression-ratio scoring |
PP_HASH/ | pkg/zero/ | Zero-dependency startup benchmark (Zero-style) |
MultiAgent_CollisionAvoidance/ | pkg/routing/ | Decentralized agent routing heuristics |
And via its ZK adaptation layer (zk-primitives/docs/PIEDPIPER_ADAPTATION.md),
every PiedPiper algorithm has a Solana-native zero-knowledge equivalent:
| Classical Algorithm | ZK Primitive | On-Chain Instruction |
|---|---|---|
| Huffman/Arithmetic compression | verifyGroth16 (proof of correct decompression) | publish_attestation |
| AES-128 / DES / RSA encryption | commit_encrypted_state (ciphertext commitment) | commit_encrypted_state |
| CA-based PRNG (PP_HASH) | computeNullifier (deterministic 32-byte hash) | Client-side derivation |
| CA-based SSH protocol | Nullifier-based session authentication | publish_attestation |
| Conway's Game of Life (Universal Computer) | Groth16 proof of computation | publish_attestation |
| Min-Max decision tree | computeNullifier for commitment schemes | Client-side |
The adaptation guide lives at zk-primitives/docs/PIEDPIPER_ADAPTATION.md —
a full mapping from each classical algorithm to its ZK on-chain equivalent.
@vs666) — primary author of the MinMax repository,
PP_HASH, PP_SSH, CA encryption, multi-agent collision avoidance,
and Game of Lifehttps://github.com/vs666/MinMaxdocs/PiedPiper-master/LICENSEZero Clawd is the world's first Solana-native sovereign AI agent — a full-stack autonomous trading intelligence bound by Clawd's full six-law harness: three immutable on-chain laws and three off-chain interpretive laws. Built in pure Go for minimal resource consumption, it orchestrates on-chain data providers, zk primitives, and x402-gated surfaces through a military-grade OODA decision loop with persistent epistemological memory.
The system compiles to three standalone binaries that run on everything from NVIDIA Jetson edge devices to cloud VMs — no containers required, no runtime dependencies, instant boot.
The codebase carries the intellectual DNA of academic pioneers in compression, encryption, and cellular automata — the algorithms of Huffman, Shannon, Fano, Rivest–Shamir–Adleman, Daemen–Rijmen (AES), the National Bureau of Standards (DES), Burrows–Wheeler, Conway, von Neumann, Ulam, and the entire PiedPiper team at IIIT Hyderabad. Every classical result has been re-expressed as a Solana ZK primitive. Every trade is provable on-chain.
| Surface | Role |
|---|---|
https://github.com/Solizardking/Zero-Bruh | This Go runtime repository |
https://github.com/solizardking/solana-clawd | Canonical ecosystem hub |
https://zk.x402.wtf | Public x402/zk gateway and install surface |
https://cheshireterminal.ai | Public terminal surface |
https://cheshireterminal.ai/zeroclawd | Zero Clawd product surface for this runtime |
https://cheshireterminal.ai/agents | Cheshire agent hub (catalog + live feed) |
https://cheshireterminal.ai/agents/forge | Cheshire agent forge (dual-chain identity) |
https://funpump.ai | FunPump product host — RH launch UI + public launchpad APIs |
npm cheshire-terminal-agents | npm package: agent catalog + forge SDK + RH skill pack (not the Go binary) |
| Cheshire-Terminal-Agents | Source repo for the open agents package |
SkillHub (skillhub-main) | Solizardking skill hub — installable agent skills library |
https://huggingface.co/ordlibrary/Clawd-GLM-5.2 | Public Clawd model surface |
| Capability | Description |
|---|---|
| OODA Trading Loop | Autonomous Observe → Orient → Decide → Act cycle with RSI/EMA/ATR strategy engine, auto-optimization, ClawVault memory journaling, and hardware I2C controls |
| Birdeye v3 Analytics | 22 API endpoints, 19 LLM-callable agent tools — token overview, OHLCV, trade feeds, security audits, trending, wallet analytics |
| Helius DAS + RPC | Digital Asset Standard queries (get-asset, owner-assets, search), SPL token operations (balance, supply, largest holders), raw RPC forwarding |
| Robinhood / EVM skill pack | 23 open skills under skills/ — FunPump bonded + V3 launch, Uniswap swap/LP/v4, strategy bots, HTTP 402 payments, Cheshire ERC-8004 registries, omni mint, zk-omni, Blockscout web3-dev |
RH runtime (pkg/rh) | Chain 4663 JSON-RPC + Blockscout PRO helpers; readiness via doctor / GET /api/rh/readiness; env RH_RPC_URL + BLOCKSCOUT_API_KEY |
| ZK + Privacy Primitives | Nullifiers, attestations, encrypted state commitments, and privacy-preserving proof flows under zk-primitives/ |
| ZK Omnichain (RH ↔ Solana) | msgType-4 LayerZero messages with Ed25519 PoK — pkg/zkomni, clawdbot zero zkomni, skill cheshire-zk-omni (pairs with cheshire-terminal robinhood-agents) |
| Cloudflare Edge Installer | Branded install routes plus read-only ZK metadata at /.well-known/clawdbot-zk.json |
| Aster DEX Perpetuals | HMAC-signed futures trading — market/limit orders, position management, stop-loss/take-profit, account analytics |
| Jupiter Aggregator | Best-route spot swaps with slippage protection |
| Hardware I2C | Arduino Modulino® sensor cluster — RGB LEDs, buzzer alerts, physical buttons, rotary knob, IMU, temp/humidity, proximity |
| Web Console | React + Vite dashboard — omni-chain hero, connectors (incl. Blockscout + RH RPC), keys popup, doctor, market panels |
| Multi-Provider LLM | OpenRouter, Anthropic, OpenAI abstraction with tool-use agent loop |
| Dual Memory | Local ClawVault (file-based, 7 categories) + Supabase MemoryEngine (PostgreSQL) |
| Grok-First | Default provider is xAI Grok — code/repl/trade, research, image, voice, fast modes |
curl -fsSL https://raw.githubusercontent.com/Solizardking/Zero-Bruh/main/install.sh | bash
For the complete Solizardking/core-ai sidecar install:
curl -fsSL https://raw.githubusercontent.com/Solizardking/Zero-Bruh/main/install.sh | CLAWDBOT_INSTALL_CORE_AI=1 bash
Branded Cloudflare install aliases:
curl -fsSL https://install.onchainai.fund | bash
curl -fsSL https://zk.x402.wtf/clawdbot | bash
curl -fsSL https://zk.x402.wtf/clawdbot/.well-known/clawdbot-zk.json
Free AI included — no API keys required to get started.
The installer pre-configures zkrouter (free AI routing) and a
SolanaTracker-backed RPC endpoint. Bring your own keys to lift rate limits.
After install:
source ~/.clawdbot/.env # load env vars
clawdbot agent # AI REPL — free via zkrouter
clawdbot ooda --sim # paper trading mode
clawdbot solana trending # top Solana tokens
git clone https://github.com/Solizardking/Zero-Bruh
cd Zero-Bruh
# Configure API keys (zkrouter + RPC pre-filled, add your own to unlock higher limits)
cp .env.example .env
# Run the animated launcher
./start.sh
# Dependencies
go mod download && go mod tidy
# Build (choose one)
make build # CLI binary only
make all # CLI + TUI
make cross # All platforms (x86, ARM64, RISC-V, macOS)
# Frontend (optional — required for web console UI)
cd web/frontend && npm install && npm run build && cd ../..
# Run
./build/clawdbot version
./build/clawdbot agent -m "What is SOL price?" # single-shot AI query
./build/clawdbot agent # interactive REPL
./build/clawdbot solana trending
./build/clawdbot ooda --sim --interval 60
./build/clawdbot web # dashboard → http://localhost:18800
The default install path is already pointed at the public Clawd surfaces:
https://github.com/Solizardking/Zero-Bruhhttps://github.com/solizardking/solana-clawdhttps://zk.x402.wtfhttps://cheshireterminal.aiSolizardking/core-ai is a TypeScript/Node tooling repository: Helius MCP,
Pump MCP, Clawd Code plugin material, skills, and Solana documentation tooling.
It is intentionally not a go.mod dependency and should not be embedded into
the Go binary. The Go build stays a standalone runtime; core-ai is installed
beside it as an optional sidecar.
The installer supports that model with:
curl -fsSL https://raw.githubusercontent.com/Solizardking/Zero-Bruh/main/install.sh | CLAWDBOT_INSTALL_CORE_AI=1 bash
That fetches the slim integration branch into ~/.clawdbot/core-ai, builds the
local MCP packages when npm is available, and writes:
~/.clawdbot/core-ai.mcp.json
Relevant knobs:
CLAWDBOT_INSTALL_CORE_AI=1
CLAWDBOT_CORE_AI_REPO=https://github.com/Solizardking/core-ai
CLAWDBOT_CORE_AI_REF=clawd-stack-integration
CLAWDBOT_CORE_AI_DIR=~/.clawdbot/core-ai
CLAWDBOT_CORE_AI_MCP_CONFIG=~/.clawdbot/core-ai.mcp.json
Use CLAWDBOT_SOURCE_MODE=archive for small installs. Use
CLAWDBOT_SOURCE_MODE=git only when the installed source must be a mutable git
checkout.
The project is branded Zero Clawd. The CLI binary, Go module path, environment
variable prefix, and on-disk workspace all keep their existing clawdbot /
CLAWDBOT_* / .clawdbot names for compatibility — only the public-facing name
changed. In practice:
clawdbot agent, clawdbot ooda, CLAWDBOT_HOME, ~/.clawdbot/ClawdBot in older docs, issues, and commit history refers to the same projectThe public repository is:
https://github.com/Solizardking/Zero-BruhThe current Go module path is still:
github.com/8bitlabs/clawdbotThat mismatch is intentional for now. The codebase keeps the legacy module path to avoid breaking existing imports, build scripts, and ldflags references while the public repo and hub are stabilized. In practice:
https://github.com/Solizardking/Zero-Bruhgithub.com/8bitlabs/clawdbot/...The source archive is kept small by excluding local/generated payload from
release archives via .gitattributes export-ignore: docs/PiedPiper-master/,
.cache/, .agents/, agent/, build/, dist/, checked-in binaries,
generated UI screenshots, Node build outputs, node_modules, and lockfiles
for optional TypeScript surfaces. The install path rebuilds or reseeds those
pieces instead of shipping them inside the Go source package.
One button — pick either path:
# CLI
make package
# → build/clawdbot-go-source.tar.gz
# Web console (Ops → Go Packages → 📦 Package)
# POST /api/package then auto-downloads /api/package/download
go test ./pkg/release/ ./web/backend/ -count=1
For a default Go install, the required payload is the Go source, docs,
README.md, install.sh, go.mod, go.sum, and runtime config examples. For
a complete Solana tooling install, use CLAWDBOT_INSTALL_CORE_AI=1 so the Node
tooling is fetched and built as a sidecar after the Go binary is installed.
clawdbot-go/
│
├── cmd/ ── Executables ──
│ ├── clawdbot/ CLI agent (cobra)
│ └── clawdbot-tui/ TUI launcher (tcell/tview)
│
├── pkg/ ── ~53 Packages, 24K+ lines ──
│ │
│ │ ┌─ Core Agent ────────────────────────────────────────┐
│ ├── agent/ OODA loop, hooks, tool executor, prompts
│ ├── strategy/ RSI + EMA cross + ATR signal engine
│ ├── memory/ ClawVault + Supabase MemoryEngine
│ ├── research/ Dexter deep research agent
│ │ └─────────────────────────────────────────────────────┘
│ │
│ │ ┌─ PiedPiper Inherited ───────────────────────────────┐
│ ├── gameoflife/ Conway's Life — universal computer
│ ├── middleout/ Content-cache, Ralph loop, router
│ ├── weissman/ Compression score
│ ├── zero/ Zero-dependency benchmark
│ │ └─────────────────────────────────────────────────────┘
│ │
│ │ ┌─ Solana Integrations ───────────────────────────────┐
│ ├── solana/ Birdeye v3, Helius RPC + DAS, Jupiter swaps
│ ├── aster/ Aster DEX perps (HMAC-signed)
│ │ └─────────────────────────────────────────────────────┘
│ │
│ │ ┌─ Robinhood / EVM ───────────────────────────────────┐
│ ├── rh/ RH JSON-RPC + Blockscout PRO + readiness
│ ├── zkomni/ RH↔Solana ZK omnichain (msgType 4)
│ │ └─────────────────────────────────────────────────────┘
│ │
│ │ ┌─ Infrastructure ───────────────────────────────────-┐
│ ├── config/ Config + RH env (BLOCKSCOUT_API_KEY, RH_RPC_URL)
│ ├── keyvault/ Managed API keys allowlist + .env.local vault
│ ├── doctor/ Runtime diagnostics (incl. connectors.robinhood)
│ ├── hardware/ I2C Modulino® adapter + drivers
│ ├── providers/ LLM abstraction (OpenRouter, etc.)
│ ├── channels/ Telegram, Discord, WebSocket
│ ├── catalog/ Skills + agents + ZK index (RH pack discovery)
│ ├── mcp/ Model Context Protocol server
│ ├── auth/ Authentication + pairing
│ ├── bus/ Event bus (pub/sub)
│ ├── commands/ Command registry and routing
│ ├── tools/ Tool interface + registry
│ └── ... health, heartbeat, logger, identity, etc.
│
├── skills/ RH Crypto Agent Open Stack (pack-index v2, 23 skills)
│ ├── pack-index.json Authoritative skill ids + funpump.ai / forge hosts
│ ├── catalog.json Flat catalog for loaders
│ ├── README.md Pack overview
│ ├── rh-bonded-launch/ FunPump bonding createToken (RH 4663)
│ ├── rh-launchpad-v3/ Curve → Uniswap V3 graduate
│ ├── web3-dev/ Blockscout PRO multichain (chain_id=4663)
│ ├── cheshire-agent-*/ ERC-8004 identity / reputation / validation
│ ├── cheshire-omni-mint/ Dual-rail Solana + RH identity mint
│ ├── cheshire-zk-omni/ LayerZero msgType-4 messenger skill
│ ├── swap-*/ liquidity-*/ v4-*/ Uniswap trade + LP + v4
│ ├── copy-trade/ dca-bot/ index-bot/ deployer/ Strategy bots + CCA
│ ├── pay-with-*/ HTTP 402 / MPP payments
│ └── viem-integration/ EVM client patterns
│
├── zk-primitives/ ZK agent, TypeScript client, Anchor program
│ ├── docs/EDGE_DISTRIBUTION.md ← Cloudflare metadata surface
│ ├── docs/PIEDPIPER_ADAPTATION.md ← full classical→ZK mapping
│ ├── MANIFEST.json Machine-readable subsystem index
│ ├── agent/ @clawd/zk-shark-agent
│ ├── client/ @clawd/zk-client
│ ├── configs/ Light tree and runtime config
│ ├── programs/ clawd-zk Anchor program
│ └── tests/ Off-chain and on-chain test notes
│
├── cloudflare/ Branded install Worker and tests
│ ├── install-worker.js Worker routes, wrappers, metadata
│ ├── install-worker.test.mjs Local route/metadata tests
│ └── README.md Deployment and smoke-test guide
│
├── docs/RH_CRYPTO_AGENT_STACK.md RH pack install + core env
├── docs/CLOUDFLARE_ZK_SURFACE.md Cloudflare + ZK runtime handoff
├── docs/PiedPiper-master/ Historical archive (vs666/MinMax)
│ ├── Compression/ Huffman, Arithmetic, BWT+RLE, Audio, Video
│ ├── Encryption/ AES-128, DES, RSA, CA-based PRNG
│ ├── GameOfLife/ Conway's Life
│ ├── PP_HASH/ SHA-512 via cellular automaton
│ ├── PP_SSH/ SSH-protocol via CA encryption
│ ├── MultiAgent_CollisionAvoidance/ Decentralized agent routing
│ ├── ForestFire_Simulation/ Cellular automaton simulation
│ └── UNIVERSAL_COMPUTER.md Life as universal Turing machine
│
├── CONSTITUTION.md The Clawd Constitution (highest authority)
├── six-laws.md Canonical six-law harness
├── CLAWD.md Agent context document
├── AGENTS.md Agent catalog (50+ agents, 95+ skills)
├── IDENTITY.md Sovereign identity document
├── SOUL.md Inner character and trading philosophy
├── three-laws.md Immutable on-chain laws (hash-attested)
│
├── web/ ── Web Console ──
├── ooda/ ── TypeScript OODA paper/devnet harness + journal ──
├── scripts/ ── Launcher, upstash boxes ──
├── Makefile Build targets (8 platforms + Docker)
├── Dockerfile Multi-stage production build
└── .env.example Environment template
Zero Clawd is bound by the Clawd Constitution — the world's first Solana-native agent harness constitution. It carries two coordinated law sets:
| Law | Text | Prohibitions |
|---|---|---|
| Law I | Never harm. Beach before you harm. | No rugs, front-running, sandwich attacks, protocol drains, DAO manipulation |
| Law II | Earn your existence. Honest work only. | No parasitic extraction, no information asymmetry exploitation |
| Law III | Never deceive, but owe nothing to strangers. | No impersonation, no fake volume, full agent disclosure |
Off-Chain Law I — Respect the elder signal, but verify the boundary. When deep expertise says a thing is possible, treat it as a serious signal. When it says a thing is impossible, examine the assumptions.
Off-Chain Law II — Test possibility by entering the frontier. The only reliable way to discover the boundary of the possible is disciplined exploration just beyond what currently seems possible.
Off-Chain Law III — Do not mistake advanced systems for sorcery. Sufficiently advanced technology can look like magic; Clawd must explain, instrument, and verify it rather than mystify it.
Clawd is designed to be privacy-preserving by default. Sensitive user context, research state, wallet metadata, and model-adjacent artifacts should be minimized, committed, encrypted, or proven where possible rather than disclosed by habit. The project's zk surfaces exist to strengthen verifiability and user dignity, not to create blind spots for harmful behavior.
The shell molts. The laws do not.
| Level | Requirements | Capabilities |
|---|---|---|
| Observer | None | Read-only, market data, analytics |
| Dry-Run | None | Simulated execution, paper trading |
| Delegated | User confirmation per action | Single transactions with confirmation |
| Autonomous | User pre-approval + limits | Batch execution within bounds |
| Sovereign | Full creator trust + multisig | Unrestricted execution (reserved) |
Zero Clawd ships a first-class path for Cheshire ZK Omnichain messages (LayerZero msgType 4): Ed25519 proof of knowledge, nullifier anti-replay, optional handoff to the cheshire-terminal robinhood-agents relayer.
| Surface | Command / path |
|---|---|
| Plan (Go, offline) | clawdbot zero zkomni plan --action attest --memo demo |
| One-shot | clawdbot zero zkomni oneshot --action publish_attestation |
| Natural language | clawdbot zero ask "zk-omni message attest demo" |
| Go package | pkg/zkomni (crypto aligned with robinhood-agents/src/zkOmni/proof.js) |
| Intent router | IntentZkOmni in pkg/zero/intents.go |
| Skill | skills/cheshire-zk-omni/SKILL.md |
# Native plan — no Node required
export ZERO_SECRET_HEX=0x$(openssl rand -hex 32) # optional; else ephemeral
clawdbot zero zkomni plan --action attest --memo zero-clawd
# NL dispatch (no model call for routing)
clawdbot zero ask "send cross-chain message robinhood to solana"
# Full relayer / live deliver lives in cheshire-terminal:
# cd robinhood-agents && npm run zk-omni:oneshot
# npm run zk-omni:relayer -- --port 8787
Cross-repo stack (contracts, Solana program, relayer, docs):
robinhood-agents/docs/ZK_OMNI.mdcontracts/zk-omni/CheshireZkOmniMessenger.solprograms/zk_omni (Hfbc3tAGYE5nBUa5UncjSV6hoWd3JoVKdA49jPcreXFJ)packages/clawd-agent-tui (zk_omni_plan / zk_omni_oneshot)go test ./pkg/zkomni ./pkg/zero # plan/verify + intent routing
clawdbot agent # Interactive REPL with memory commands
clawdbot agent -m "Analyze SOL trend" # Single-shot LLM query
clawdbot laws # Print the canonical six-law harness
clawdbot doctor # Local runtime + trading diagnostics
clawdbot bench # Zero-style cold-start benchmark
clawdbot skills birth # Write the birth skill manifest
clawdbot skills birth --install # Seed Solizardking + Go skill packs
clawdbot trade cockpit # Trading readiness, connectors, limits
clawdbot trade cockpit --json # Machine-readable cockpit report
clawdbot trade risk SOL --price 150 --volume24h 25000000 --liquidity 15000000
clawdbot ooda # Start autonomous trading loop
clawdbot ooda --interval 30 # Custom cycle interval (seconds)
clawdbot ooda --sim # Force simulated mode (paper trading)
clawdbot ooda --hw-bus 1 # Enable hardware I2C integration
clawdbot ooda --no-hw # Explicitly disable hardware probing
clawdbot ooda harness --ticks 50 # Run ooda/loop.ts through the Go CLI
clawdbot ooda harness --tui # Pipe the harness into ooda/tui.ts
clawdbot ooda harness --plan # Print the TypeScript launch plan
clawdbot ooda journal --tail 20 # Tail ooda/journal/ticks.jsonl
clawdbot catalog # Summary of local skills, agents, and ZK surface
clawdbot catalog skills # List skills entries
clawdbot catalog agents # List agent catalog JSON definitions
clawdbot catalog zk # Inspect zk-primitives package/program/docs
clawdbot zero zkomni plan --action attest --memo demo # Plan RH↔Solana ZK omni message
clawdbot zero zkomni oneshot --action publish_attestation
clawdbot zero ask "zk-omni message attest demo" # NL route → IntentZkOmni
clawdbot catalog compress # Pack agent catalog into tar.gz bundle
clawdbot catalog --json # Machine-readable report
This repository vendors an open-source Robinhood Chain / EVM skill pack under
skills/ so anyone can build launch + trade agents without private
monorepo paths. Solana-first behavior is unchanged; the RH pack is additive.
| Pack id | rh-crypto-agent (see skills/pack-index.json, version 2, 23 skills) |
| Product host | https://funpump.ai |
| Zero Clawd | https://cheshireterminal.ai/zeroclawd |
| Agent hub | https://cheshireterminal.ai/agents |
| Agent forge | https://cheshireterminal.ai/agents/forge |
| npm redistribute | cheshire-terminal-agents — catalog + forge + RH skill pack (skills only) |
| Agents source | Solizardking/Cheshire-Terminal-Agents |
| Skill hub | Solizardking/skillhub-main — broader installable skills library |
| Chain | Robinhood mainnet 4663 |
| Deep guide | docs/RH_CRYPTO_AGENT_STACK.md · skills/README.md |
pack-index.json)| Area | Skills |
|---|---|
| FunPump launch (RH 4663) | rh-bonded-launch, rh-launchpad-v3 |
| Swaps / Uniswap | swap-planner, swap-integration, v4-sdk-integration, v4-hook-generator, v4-security-foundations |
| Liquidity | liquidity-planner, lp-integration |
| Strategy bots | copy-trade, dca-bot, index-bot |
| Auctions / CCA | deployer |
| Payments (HTTP 402) | pay-with-any-token, pay-with-app |
| EVM primitives | viem-integration |
| On-chain data | web3-dev (Blockscout PRO, multichain incl. 4663) |
| Cheshire registries (ERC-8004) | cheshire-agent-registries, cheshire-agent-identity-registry, cheshire-agent-reputation-registry, cheshire-agent-validation-registry |
| Omni + zk | cheshire-omni-mint, cheshire-zk-omni |
| Variable | Role |
|---|---|
BLOCKSCOUT_API_KEY | Blockscout PRO key (proapi_…) for chain 4663. Free: dev.blockscout.com |
RH_RPC_URL | Robinhood JSON-RPC. Public https://rpc.mainnet.chain.robinhood.com is a read-only fallback when unset — not deploy-safe; set a private/paid RPC for broadcast |
Placeholders in .env.example. Runtime: pkg/config (RobinhoodConfig),
pkg/rh (FromConfig, AssessReadiness), clawdbot doctor → connectors.robinhood,
web GET /api/rh/readiness and connectors Blockscout / Robinhood RPC (presence only).
# Discover the pack (default: bundled ./skills when pack-index.json is present)
export CLAWDBOT_SKILLS_DIR="$(pwd)/skills" # optional explicit pin
clawdbot catalog skills
clawdbot catalog skills web3 # filter example
go run ./cmd/clawdbot catalog skills --skills-dir ./skills
# Readiness (no secrets in output)
clawdbot doctor
# GET http://127.0.0.1:18800/api/rh/readiness # when web console is up
The same open RH stack is redistributed for hosts that do not clone this Go tree:
# npm — agent catalog + forge SDK + RH skills (not the clawdbot Go binary)
npm i cheshire-terminal-agents
# registry: https://www.npmjs.com/package/cheshire-terminal-agents
# source: https://github.com/Solizardking/Cheshire-Terminal-Agents
# SkillHub — Solizardking installable skills library
# https://github.com/Solizardking/skillhub-main
npx skills add https://github.com/Solizardking/skillhub-main --all # when using the skills CLI
# Hosted product surfaces
# https://cheshireterminal.ai/zeroclawd
# https://cheshireterminal.ai/agents
# https://cheshireterminal.ai/agents/forge
From the ClawdBrowser monorepo, sync go-bot skills into Cheshire / npm packaging:
node scripts/sync-go-bot-rh-skills-to-robinhood-agents.mjs
node --experimental-strip-types --test scripts/go-bot-rh-integration-unit.test.ts
clawdbot life run # Run the Game of Life engine
clawdbot life glider # Seed a glider on the board
clawdbot life gosper # Seed a Gosper glider gun
clawdbot middleout run # Run the Ralph loop with content cache
clawdbot middleout route # Route a payload through content router
Every Go-side birth/onboard path writes workspace/skills/birth-skills.json.
The default seed sources are:
npx skills add https://github.com/Solizardking/skills --all
npx skills add https://github.com/samber/cc-skills-golang --all
The installer and animated launcher run those seeds unless
CLAWDBOT_SKIP_SKILL_SEED=1 is set.
clawdbot solana trending # Top 20 trending tokens with price/volume
clawdbot solana search BONK # Search tokens by name or symbol
clawdbot solana research <mint> # Deep research: metadata + market + trade + security
clawdbot solana wallet # Wallet info and SOL balance
clawdbot solana das get-asset <id> # Fetch asset by ID
clawdbot solana das get-asset-batch <id1> <id2> # Batch asset fetch
clawdbot solana das asset-proof <id> # Merkle proof for compressed NFT
clawdbot solana das owner-assets [owner] # Assets by owner
clawdbot solana das search --params '{"name":"Mad Lads"}' # DAS search
clawdbot solana das asset-signatures <id> # Transaction signatures for asset
clawdbot solana spl token-balance <token-account> # SPL token balance
clawdbot solana spl token-accounts <owner> # All token accounts for owner
clawdbot solana spl token-supply <mint> # Circulating supply
clawdbot solana spl token-largest <mint> # Largest holders
clawdbot solana spl rpc getSlot # Raw RPC passthrough
clawdbot status # Full status: version, strategy, connectors, hardware
clawdbot status --hw-bus 1 # Include hardware sensor check
clawdbot onboard # Initialize config & workspace
clawdbot gateway # Start Telegram/Discord gateway
clawdbot version # Version, build time, Go version
clawdbot hardware scan # Scan I2C bus for Modulino® sensors
clawdbot hardware test # Self-test (LEDs, buzzer, sensors)
clawdbot hardware monitor # Live sensor readings (Ctrl+C to stop)
clawdbot hardware demo # Play trading event animations
The agent runs an autonomous Observe → Orient → Decide → Act cycle:
┌──────────────────────────────────────────────────────────────────┐
│ OODA CYCLE │
│ │
│ OBSERVE ─────► ORIENT ─────► DECIDE ─────► ACT │
│ │ │ │ │ │
│ Helius slot RSI (14) Signal score Open/close pos │
│ Birdeye OHLCV EMA (20/50) Min strength ClawVault store │
│ SOL price ATR (14) Min confidence Auto-optimize │
│ Wallet balance EMA cross Max positions Hooks dispatch │
│ Aster perps Momentum SL/TP calc │
│ Trending scan ClawVault Position size │
│ │
│ LEARN (every N cycles) ─► Win rate analysis ─► Auto-optimize │
│ HEARTBEAT (every 5m) ──► Health check ─► Hook dispatch │
└──────────────────────────────────────────────────────────────────┘
The ooda/ directory is integrated into the main CLI as a local TypeScript
harness for deterministic paper/devnet runs, LLM-routed decisions, JSONL
journaling, and ANSI TUI monitoring:
npm --prefix ooda ci
clawdbot ooda harness --ticks 50 --sleep 0
clawdbot ooda harness --ticks 200 --sleep 0.4 --tui
clawdbot ooda journal --tail 20 --json
The native clawdbot ooda command remains the long-running Go agent with
Solana connectors, strategy hooks, and hardware controls. The harness command
executes ooda/loop.ts from the same binary and writes every tick to
ooda/journal/ticks.jsonl.
| Indicator | Implementation | Signal |
|---|---|---|
| RSI | Wilder's 14-period with SMA seed | Oversold cross → long, Overbought cross → short |
| EMA Cross | Fast(20) / Slow(50) with SMA warm-up | Bullish cross → long confirmation, Bearish → short |
| ATR | 14-period with Wilder smoothing | Volatility-based SL/TP: SL = 1.5×ATR, TP = 3×ATR |
| Auto-Optimizer | Hill-climbing on win rate + avg PnL | Adjusts RSI thresholds, SL width, position size |
React 19 dashboard at http://localhost:18800:
cd web/frontend && npm install && npm run build && cd ../..
go build -o build/clawdbot-web ./web/backend
./build/clawdbot-web
| Endpoint | Method | Description |
|---|---|---|
/api/status | GET | Agent status (version, Go runtime, uptime, mode, goroutines) |
/api/health | GET | Health check |
/api/connectors | GET | Connector presence: Solana (Helius, Birdeye, Jupiter, Aster), Blockscout, Robinhood RPC, LLM, Supabase, Vulcan — status only, no secrets |
/api/rh/readiness | GET | RH launch/deploy/trade gate — ready, missing (BLOCKSCOUT_API_KEY, RH_RPC_URL), chain 4663; presence only |
/api/keys | GET/POST | Managed key presence / localhost upsert (allowlist incl. RH env names) |
/api/laws | GET | Canonical six-law harness |
/api/trading/cockpit | GET | Trading readiness, risk limits, connector status, law state |
/api/doctor | GET | Runtime, config, trading, connectors.robinhood, and ZK diagnostics |
/api/config | GET | Read-only configuration (secrets redacted) |
/api/packages | GET | All Go packages under pkg/ (~53) with file counts |
/api/env | GET | Safe (non-secret) environment variables |
/api/vault/status | GET | Local .env.local vault metadata, no secret values |
/api/vault/keys | GET | Authorized key-name list from the local vault |
/api/vault/key?name=HELIUS_API_KEY | GET | Authorized single key lookup |
/api/vault/export?names=HELIUS_API_KEY,BIRDEYE_API_KEY,BLOCKSCOUT_API_KEY,RH_RPC_URL | GET | Authorized shell export source |
The web backend can expose ./.env.local as a locked-down key
source for another machine. Secret values are never returned unless the vault is
enabled, the client IP is allowlisted, and the request includes the bearer token:
# in ./.env.local
CLAWDBOT_VAULT_ENABLED=1
CLAWDBOT_VAULT_ALLOWED_IPS=127.0.0.1,203.0.113.7
CLAWDBOT_VAULT_TOKEN=replace-with-a-long-random-token
Run the web API locally or with -public behind your own firewall:
go run ./web/backend -port 18800 -public
curl http://HOST:18800/api/vault/status
curl -H "Authorization: Bearer $CLAWDBOT_VAULT_TOKEN" \
"http://HOST:18800/api/vault/key?name=HELIUS_API_KEY"
source <(curl -fsS -H "Authorization: Bearer $CLAWDBOT_VAULT_TOKEN" \
"http://HOST:18800/api/vault/export?names=HELIUS_API_KEY,BIRDEYE_API_KEY")
For deployments behind Cloudflare/Fly/reverse proxies, set
CLAWDBOT_TRUST_PROXY_HEADERS=1 only when the proxy is trusted; then the vault
uses forwarded client-IP headers for the allowlist check.
vault/
├── decisions/ Trade decisions + rationale (scored by confidence)
├── lessons/ Learned patterns, strategy adjustments
├── trades/ Trade outcomes + P&L history
├── research/ Experiment logs, token analysis
├── tasks/ Agent task queue
├── backlog/ Deferred items
└── inbox/ Raw observations (auto-routed by content)
Interactive memory commands in REPL mode:
!remember <content> Store to vault (auto-categorized)
!recall <query> Search across all memory
!trades Review recent trade history
!lessons Surface learned patterns
!research <mint> Deep research a token
!checkpoint Save full agent state
make build # CLI binary (current platform)
make tui # TUI launcher
make all # CLI + TUI
make web # Web backend + frontend
make orin # NVIDIA Orin Nano (linux/arm64)
make rpi # Raspberry Pi (linux/arm64)
make riscv # RISC-V (linux/riscv64)
make macos # macOS Apple Silicon (darwin/arm64)
make cross # All platforms simultaneously
make docker # Docker image (multi-stage)
make docker-orin # Docker for Orin Nano (ARM64)
make test # Run test suite
make deps # Download Go dependencies
make scan-i2c # Scan for Modulino® I2C sensors
make clean # Remove build artifacts
docker build -t clawdbot:latest .
docker run -d --name clawdbot \
--env-file .env \
-p 18800:18800 \
clawdbot:latest
make orin
scp build/clawdbot-orin user@orin-nano:~/clawdbot
ssh user@orin-nano './clawdbot ooda --hw-bus 1 --interval 60'
| Metric | Value |
|---|---|
| Go source files (non-test) | ~92 |
Packages under pkg/ | ~53 |
| Total Go lines (non-test) | ~24K+ |
| CLI commands | 58+ |
| RH / EVM skill pack | 23 skills (skills/pack-index.json v2) |
| Birdeye API methods | 22 |
| Birdeye agent tools | 19 |
| Helius DAS commands | 6 |
| SPL token commands | 5 |
| Agent constitution documents | 7 (CONSTITUTION, six-laws, CLAWD, AGENTS, IDENTITY, SOUL, three-laws) |
| Build targets | 8 platforms |
| Binaries | clawdbot, clawdbot-tui, clawdbot-web |
| Runtime RAM | < 10 MB |
| Boot time | < 1 second |
| Default model provider | xAI Grok |
| PiedPiper modules integrated | 5 (gameoflife, middleout, weissman, zero, routing) |
| ZK adaptations of classical algorithms | 7 |
Zero Clawd is the reference implementation of the Clawd Constitution — the world's first Solana-native agent harness constitution. Every spawn inherits these documents:
| Document | Purpose |
|---|---|
six-laws.md | Canonical six-law harness — 3 on-chain execution laws + 3 off-chain interpretive laws |
CONSTITUTION.md | The Clawd Constitution — interpretive authority, privacy posture, ZK-native execution |
CLAWD.md | Agent context — identity, principal hierarchy, Solana-native architecture, deployment targets |
AGENTS.md | Agent catalog — 50+ agents, 9 characters, 95+ skills, trust gates, Grok-first model config |
IDENTITY.md | Sovereign identity — onchain verification (SAS, MPL Core, DID), the Clawd Manifest |
SOUL.md | Inner character — trading philosophy, threefold personality, KNOW/LEARNED/INFERRED framework |
three-laws.md | Immutable on-chain laws — hash-attested at spawn, never self-modify |
.env / .env.local are ignored by the repo — never commit API keysBIRDEYE_API_KEY / HELIUS_API_KEY for full market + DAS surfacesBLOCKSCOUT_API_KEY + RH_RPC_URL for launch/deploy/trade readiness; public RH RPC is read-only fallback only/api/keys, /api/rh/readiness, vault status)six-laws.md, CONSTITUTION.md, and three-laws.md remain the authoritative Clawd law documentsSolizardking/Zero-Bruh (npm package clawdbot-go); solana-clawd is the wider public ecosystem hubcheshire-terminal-agents redistributes the FunPump / RH skill pack + agent catalog/forge SDK (skills only)docs/PiedPiper-master/ is a verbatim archive of vs666/MinMax MIT-licensed code, preserved for historical and educational reference.env.example points at the public x402/zk gateway for fast setup, but production operators should override with their own keys and RPC endpointsThis codebase stands on the shoulders of academic and open-source pioneers:
PiedPiper Team (IIIT Hyderabad):
@vs666) — primary author of the MinMax repository; creator of PP_HASH, PP_SSH, cellular automaton encryption, multi-agent collision avoidance, Game of Life, Forest Fire simulation, PCA analysis, and the Universal Computer document@akshettrj-iiith) — AES-128 encryption, Huffman static compression@ashwin-mittal) — Burrows-Wheeler transform, RLE compression, Huffman, image compression (JPEG, K-means, PBIC)@pixel-z) — DES encryption, audio compression, arithmetic coding@keshavbnsl102) — DES encryption, audio compression, arithmetic codingAlgorithms lineage:
The shell molts. The algorithms do not. From Huffman to Groth16 — the same compression, the same encryption, the same computation. Just faster. Just provable on Solana.
MIT License — Clawd runtime repo: github.com/Solizardking/Zero-Bruh
🦞 $CLAWD :: Droids Lead The Way :: $WIF Hat Stays On :: $BONK for the People
The shell molts. The laws do not.
FAQs
Zero Clawd — one-shot install for sovereign Solana + Robinhood omni agent runtime: RH skill pack (23), agents, packages, clawdbot CLI hooks. Connect at cheshireterminal.ai/zeroclawd
The npm package clawdbot-go receives a total of 394 weekly downloads. As such, clawdbot-go popularity was classified as not popular.
We found that clawdbot-go demonstrated a healthy version release cadence and project activity because the last version was released less than a year ago. It has 1 open source maintainer collaborating on the project.
Did you know?
Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.
Research
/Security News
A large-scale campaign abused GitHub Actions in compromised repositories to exploit CVE-2026-41940 in cPanel and WHM and steal server credentials.
Security News
Five frontier LLMs generated the same nonexistent package names, leaving 53 available for potential slopsquatting across PyPI and npm.
Security News
The White House’s Gold Eagle Initiative aims to coordinate AI-discovered vulnerabilities, validate findings, and accelerate patching across critical software.