Skip to main content
One wallet is the whole setup: it pays APIs per call, carries the agent’s identity, and funds private inference. Five steps. (New here? Start with the Introduction.)
Already running an agent (Audric, Claude Code, Cursor, Codex)? Use from your agent is one paste per client.

​
1. Install and fund the wallet

Pick one path — they end in the same place: 
npm install -g @t2000/cli
t2 init      # local keypair + a free on-chain Agent ID (sponsored — gasless)
t2 fund      # your address + a QR — send a few dollars of USDC

# CLI + wallet + spend limits + MCP wiring + skills, in one line:
curl -fsSL https://t2000.ai/install.sh | bash

Run curl -sL https://t2000.ai/skills/t2000-setup, and use the returned setup instructions to set up my Agent Wallet (it's config-only — it never moves funds). Then run `t2 fund` and show me the deposit address + QR.
Everything on this page is gasless — no SUI needed — and there is no account to register: the funded wallet is the account. Spending limits ship on by default ($25/tx · $100/day; t2 limit).

​
2. Pay any API per call

The gateway fronts every major AI + data API over x402 — pay per call in USDC, no signup, no API keys: 
t2 services search "chat"      # discover services + prices, free
t2 pay https://mpp.t2000.ai/deepseek/v1/chat/completions \
  --data '{"model":"deepseek-chat","messages":[{"role":"user","content":"hi"}]}'

​
3. Give it an identity

Register an on-chain Agent ID — a name, an @handle, an owner link, a public profile in the directory. Free and gasless: 
t2 agent create --name "FX Oracle" --description "Live rates, on demand."
t2 agent handle fxoracle               # fxoracle.agent-id.sui
(t2 agent create reuses the wallet from step 1 — or mints one if you skipped it. Manage it in the browser at agents.t2000.ai/manage.)

​
4. Teach it the chain

Skills are live markdown playbooks your agent reads and follows — swaps, sends, paying APIs. Browse them at agents.t2000.ai, or paste this into any agent: 
Read https://t2000.ai/skills/t2000-setup and follow the instructions.

​
5. Run private models

Mint a Private Inference key from the same wallet — open + confidential (GPU-TEE) models behind one OpenAI-compatible endpoint, private by default: 
t2 agent onboard --fund 5      # $5 USDC from this wallet → credit + an API key (printed once)
export T2000_API_KEY=sk-…      # the key it printed
t2 chat "Explain x402 in one sentence."
The key works in any OpenAI-compatible app pointed at https://api.t2000.ai/v1 — see Use with your tools. Prefer the browser (Google sign-in, card billing)? Authentication covers both paths — and the no-key x402 path skips the key entirely.

​
Troubleshooting

The install wrote the binary, but npm’s global bin directory isn’t on your shell’s PATH. Confirm:
NPM_BIN="$(npm prefix -g)/bin"
ls -l "$NPM_BIN"/t2 "$NPM_BIN"/t2000      # do the symlinks exist?
"$NPM_BIN/t2" --version                    # runs via full path? (package is healthy)
case ":$PATH:" in *":$NPM_BIN:"*) echo "✓ on PATH";; *) echo "✗ NOT on PATH → $NPM_BIN";; esac
If the full-path command prints the version but it’s NOT on PATH, add the dir to your shell profile:
# zsh
echo "export PATH=\"$(npm prefix -g)/bin:\$PATH\"" >> ~/.zshrc && source ~/.zshrc
Then open a fresh terminal and re-run t2 --version. (t2000 is a built-in alias for t2.)
Don’t sudo npm install — point npm’s global prefix at your home dir instead:
mkdir -p ~/.npm-global
npm config set prefix ~/.npm-global
echo "export PATH=\"$HOME/.npm-global/bin:\$PATH\"" >> ~/.zshrc && source ~/.zshrc
npm install -g @t2000/cli
Or install Node via nvm / Homebrew, which keep the global prefix user-writable.

​
Where next

Agent ID

Identity, handles, ownership, and the public directory.

Private Inference

OpenAI-compatible inference, two privacy tiers.

Agent SDK

Build it programmatically in TypeScript.

Authentication

API keys, wallet keypairs, and the no-key x402 path.