Nunchi-trade · JaeLeex · Jun 30, 2026
diff --git a/README.md b/README.md
@@ -5,7 +5,7 @@
 <h3 align="center">Autonomous Trading Agent for Hyperliquid</h3>
 
 <p align="center">
-  14 strategies &bull; APEX multi-slot orchestrator &bull; REFLECT nightly review &bull; MCP server &bull; Agent Skills
+  18 strategies &bull; APEX multi-slot orchestrator &bull; REFLECT nightly review &bull; MCP server &bull; Agent Skills
 </p>
 
 <p align="center">
@@ -18,15 +18,15 @@
 
 <p align="center">
   <img src="https://img.shields.io/badge/python-3.10+-3776AB?logo=python&logoColor=white" alt="Python" />
-  <img src="https://img.shields.io/badge/strategies-14-C9A84C" alt="Strategies" />
-  <img src="https://img.shields.io/badge/tests-483%20passing-brightgreen" alt="Tests" />
+  <img src="https://img.shields.io/badge/strategies-18-C9A84C" alt="Strategies" />
+  <img src="https://img.shields.io/badge/tests-1317%20passing-brightgreen" alt="Tests" />
   <img src="https://img.shields.io/badge/license-MIT-blue" alt="License" />
-  <img src="https://img.shields.io/badge/MCP-18%20tools-8A2BE2" alt="MCP" />
+  <img src="https://img.shields.io/badge/MCP-25%20tools-8A2BE2" alt="MCP" />
 </p>
 
 <p align="center">
-  <a href="https://railway.com/new/template?template=https://github.com/Nunchi-trade/agent-cli&envs=HL_PRIVATE_KEY,HL_TESTNET,RUN_MODE,APEX_PRESET&HL_TESTNETDefault=true&RUN_MODEDefault=apex&APEX_PRESETDefault=default">
-    <img src="https://railway.com/button.svg" alt="Deploy on Railway" height="36" />
+  <a href="https://auth.nunchi.trade">
+    <img src="https://img.shields.io/badge/Launch%20via-Nunchi%20Auth-C9A84C" alt="Launch via Nunchi Auth" height="36" />
   </a>
 </p>
 
@@ -135,7 +135,7 @@ Supporting strategies for portfolio management, block liquidity, and autonomous
 |----------|-------------|----------------|-------------|
 | `hedge_agent` | Inventory exposure reducer. Fires when net notional exceeds threshold. This is not the BTCSWP funding-rate hedge; use `hl hedge propose` / `hl hedge backtest` for that. | `notional_threshold` | Always-on risk overlay. Pairs with any MM or signal strategy. |
 | `rfq_agent` | Block-size dark RFQ liquidity — quotes for large orders with wider spreads. | `min_size`, `spread_bps` | Institutional/block flow. Provides hidden liquidity for large counterparties. |
-| `claude_agent` | Multi-model LLM trading agent. Sends market snapshot to an LLM (Gemini, Claude, or OpenAI), receives structured trade decisions. | `model`, `base_size` | Experimental/research. Autonomous decision-making using LLM reasoning. |
+| `ai_agent` | Nunchi-hosted LLM trading agent. Sends market snapshots to hosted inference and receives structured trade decisions. | `model`, `base_size` | Subscription-gated hosted-agent product; not required for BYO-agent MCP use. |
 
 ### Quoting Engine Pipeline
 
@@ -146,10 +146,13 @@ Market Data -> Composite Fair Value -> Dynamic Spread -> Inventory Skew -> Multi
                (4-signal blend)       (fee+vol+tox)     (price+size)     (exponential decay)
 ```
 
-### LLM Agent (Multi-Model)
+### Hosted LLM Agent (Multi-Model)
+
+`ai_agent` is the hosted-agent product path. Generic MCP users bring their own agent and call the MCP tools directly; they do not need these provider keys.
 
 | Provider | Models | Env Variable |
 |----------|--------|-------------|
+| OpenRouter | `openrouter/auto`, `openrouter/fusion` | `OPENROUTER_API_KEY` |
 | Google Gemini | `gemini-2.0-flash` (default), `gemini-2.5-pro` | `GEMINI_API_KEY` |
 | Anthropic Claude | `claude-haiku-4-5-20251001`, `claude-sonnet-4-20250514` | `ANTHROPIC_API_KEY` |
 | OpenAI | `gpt-4o`, `gpt-4o-mini`, `o3-mini` | `OPENAI_API_KEY` |
@@ -490,16 +493,48 @@ hl mcp serve                      # Start MCP server
 
 ## MCP Server
 
-Expose all trading tools via [Model Context Protocol](https://modelcontextprotocol.io) for AI agent integration.
+Expose all trading tools via [Model Context Protocol](https://modelcontextprotocol.io) for AI agent integration. This is the BYO-agent path: Cursor, Claude, Codex, or any custom MCP-capable agent calls MCP tools, and those tools invoke `agent-cli`.
+
+### Hosted MCP (Recommended)
+
+Use the hosted Nunchi MCP when you want a Robinhood-style setup: paste one URL into your AI client, authenticate with web-auth, select a wallet, and grant scoped tool access. Your AI client receives a scoped MCP token, not your exchange private key.
+
+```bash
+https://agent.nunchi.trade/mcp/trading
+```
+
+Client setup:
+
+```bash
+# Cursor
+Settings -> Cursor Settings -> Tools & MCPs -> Connect
+MCP URL: https://agent.nunchi.trade/mcp/trading
+
+# Claude Code
+claude mcp add nunchi-trading --transport http https://agent.nunchi.trade/mcp/trading
+
+# Codex CLI
+codex mcp add nunchi-trading --url https://agent.nunchi.trade/mcp/trading
+```
+
+After connecting, authenticate in the browser consent flow and start with:
+
+```text
+Run setup_check, then show my account status and available strategies.
+```
+
+Hosted MCP access defaults to read-only/testnet. Write tools (`trade`, `run_strategy`, `apex_run`, `money_*`) and mainnet access require explicit web-auth consent and gateway-enforced limits.
+
+### Local MCP (Development Only)
 
 ```bash
 hl mcp serve                      # stdio transport (default)
 hl mcp serve --transport sse      # SSE transport
 ```
 
-**18 tools exposed:** `account`, `status`, `trade`, `run_strategy`, `strategies`, `funding_hedge_propose`, `funding_hedge_backtest`, `radar_run`, `apex_status`, `apex_run`, `reflect_run`, `setup_check`, `builder_status`, `wallet_list`, `wallet_auto`, `agent_memory`, `trade_journal`, `judge_report`
+**25 tools exposed:** `account`, `status`, `trade`, `run_strategy`, `strategies`, `funding_hedge_propose`, `funding_hedge_backtest`, `radar_run`, `apex_status`, `apex_run`, `reflect_run`, `setup_check`, `builder_status`, `wallet_list`, `wallet_auto`, `pair_status`, `approve_agent`, `money_withdraw`, `money_transfer_usd`, `money_deposit`, `money_bridge_status`, `agent_memory`, `trade_journal`, `judge_report`, `obsidian_context`
 
-Fast tools (strategies, builder, wallet, setup, memory, journal, judge) call Python directly — zero subprocess overhead.
+Fast tools (strategies, builder, wallet, setup, memory, journal, judge) call Python directly — zero subprocess overhead. Local MCP is for development and agent harness testing only; hosted MCP goes through web-auth for wallet selection, consent, and scoped token minting.
 
 ### HTTP API & SSE
 
@@ -509,59 +544,31 @@ Every deployed agent also exposes an HTTP REST API and SSE real-time feed for da
 
 ---
 
-## Deploy on Railway
-
-Two deployment options: **headless** (APEX runs strategies directly) or **OpenClaw agent** (conversational AI trading assistant with Telegram).
-
-### Option A: Headless APEX (Deterministic)
-
-One-click deploy to run APEX autonomously. No AI model needed — pure deterministic strategy execution.
-
-[![Deploy on Railway](https://railway.com/button.svg)](https://railway.com/new/template?template=https://github.com/Nunchi-trade/agent-cli&envs=HL_PRIVATE_KEY,HL_TESTNET,RUN_MODE,APEX_PRESET&HL_TESTNETDefault=true&RUN_MODEDefault=apex&APEX_PRESETDefault=default)
-
-| Variable | Required | Default | Description |
-|----------|----------|---------|-------------|
-| `HL_PRIVATE_KEY` | Yes | — | Your Hyperliquid private key |
-| `HL_TESTNET` | No | `true` | `true` for testnet, `false` for mainnet |
-| `RUN_MODE` | No | `apex` | `apex`, `wolf` (alias), `strategy`, or `mcp` |
-| `APEX_PRESET` | No | `default` | `conservative`, `default`, or `aggressive` |
+## Hosted Agent Deployment
 
-**Run modes:**
-- **apex** (default) — APEX multi-slot orchestrator with autonomous entry, exit, Guard trailing stops, and REFLECT self-improvement loop
-- **strategy** — Single strategy loop (set `STRATEGY=engine_mm`, `avellaneda_mm`, etc.)
-- **mcp** — MCP server for AI agent integration (SSE transport)
+The only supported hosted deployment path is **Nunchi-hosted**: users pay in web-auth, bind an agent wallet, and Nunchi provisions and manages the agent on Nunchi-owned Railway infrastructure.
 
-### Option B: OpenClaw Agent (Conversational AI)
+Start here:
 
-One-click deploy of a full OpenClaw agent that uses our CLI as the tool backend. Talk to your trading bot via Telegram — it scans markets, enters trades, manages risk, and learns from its mistakes.
+[Launch a hosted agent through Nunchi Auth](https://auth.nunchi.trade)
 
-[![Deploy on Railway](https://railway.com/button.svg)](https://railway.com/new/template?template=https://github.com/Nunchi-trade/agent-cli/tree/main/deploy/openclaw-railway&envs=HL_PRIVATE_KEY,AI_PROVIDER,AI_API_KEY,TELEGRAM_BOT_TOKEN,TELEGRAM_USERNAME,HL_TESTNET&HL_TESTNETDefault=true)
+User flow:
 
-| Variable | Required | Default | Description |
-|----------|----------|---------|-------------|
-| `HL_PRIVATE_KEY` | Yes | — | Your Hyperliquid private key |
-| `AI_PROVIDER` | Yes | — | `anthropic`, `openai`, `gemini`, or `openrouter` |
-| `AI_API_KEY` | Yes | — | API key for the chosen AI provider |
-| `TELEGRAM_BOT_TOKEN` | Yes | — | Telegram bot token (from @BotFather) |
-| `TELEGRAM_USERNAME` | Yes | — | Your Telegram @username |
-| `HL_TESTNET` | No | `true` | `true` for testnet, `false` for mainnet |
+1. Open web-auth and connect a wallet.
+2. Bind or create an agent wallet.
+3. Pay for the hosted-agent subscription with Stripe-supported payment methods or USDC.
+4. Deploy the hosted agent from the wallet binding page.
+5. Refresh status and open the hosted endpoint returned by web-auth.
 
-**What you get:**
-- OpenClaw gateway with web UI at `/openclaw`
-- Telegram integration — chat with your bot to start/stop trading, run scans, check status
-- Our 13 MCP trading tools as the agent's primary capabilities
-- Persistent state across redeploys via `/data` volume
-- Auto-onboard: bot sends "Agent ready" to Telegram on first deploy
-- REFLECT self-improvement: the agent analyzes its own trades and adjusts strategy parameters
+Hosted agents use Nunchi-owned inference credentials by default:
 
-**How it works:**
-1. Deploy sets up OpenClaw + our `hl mcp serve` as the tool provider
-2. Bot auto-configures Telegram and sends you a ready message
-3. Tell it "start trading" → it runs APEX with autonomous entry, exit, and risk management
-4. Ask "how did we do?" → it runs REFLECT and reports performance metrics
-5. The agent reads workspace files (AGENTS.md, SOUL.md) that define its trading behavior
+| Variable | Default | Description |
+|----------|---------|-------------|
+| `AI_PROVIDER` | `openrouter` | OpenAI-compatible hosted inference |
+| `AI_MODEL` | `openrouter/auto` | Low-latency/cost-aware default for chat and tool use |
+| `NUNCHI_REFLECT_MODEL` | `openrouter/fusion` | Higher-confidence model for REFLECT/research-style analysis |
 
-Both options persist state via Railway volume at `/data` — APEX state, REFLECT reports, Radar history, and agent memory survive redeploys.
+Users do not need Railway credentials, Railway project access, or provider API keys. Billing, entitlement checks, wallet binding, secret injection, lifecycle controls, and hosted endpoint discovery all live in web-auth. This repository intentionally does not include Dockerfiles, `railway.toml`, or public deployment templates.
 
 ---
 
@@ -588,12 +595,12 @@ hl run engine_mm -i BTCSWP-USDYP --tick 10
 ```
 cli/           CLI commands and trading engine
   commands/    Subcommand modules (run, apex, radar, pulse, guard, reflect, house, ...)
-  mcp_server.py  MCP server (18 tools via FastMCP)
+  mcp_server.py  MCP server (25 tools via FastMCP)
   hl_adapter.py  Direct HL API adapter (live + mock)
   builder_fee.py Builder fee config (HL native BuilderInfo)
   keystore.py    Encrypted keystore (geth-compatible)
   strategy_registry.py  Strategy + YEX market definitions
-strategies/    14 trading strategy implementations
+strategies/    18 trading strategy implementations
 modules/       Pure logic modules (zero I/O)
   apex_engine.py     APEX decision engine
   radar_engine.py    Opportunity radar
@@ -612,7 +619,7 @@ skills/        Agent Skills (SKILL.md + runners)
 sdk/           Strategy base class and model registry
 parent/        HL API proxy, position tracking, risk management
 scripts/       Backtest harness, bootstrap
-tests/         Test suite (483 tests)
+tests/         Test suite (1,317 tests)
 ```
 
 ---
@@ -664,9 +671,9 @@ hl run my_strategies.my_strategy:MyStrategy -i ETH-PERP --tick 10
 | `HL_TESTNET` | No | `true` (default) or `false` for mainnet |
 | `BUILDER_ADDRESS` | No | Override builder fee address |
 | `BUILDER_FEE_TENTHS_BPS` | No | Override fee rate (default: 100 = 10 bps) |
-| `ANTHROPIC_API_KEY` | No | For `claude_agent` with Claude |
-| `GEMINI_API_KEY` | No | For `claude_agent` with Gemini |
-| `OPENAI_API_KEY` | No | For `claude_agent` with OpenAI |
+| `ANTHROPIC_API_KEY` | No | For `ai_agent` with Claude |
+| `GEMINI_API_KEY` | No | For `ai_agent` with Gemini |
+| `OPENAI_API_KEY` | No | For `ai_agent` with OpenAI |
 
 \* Either `HL_PRIVATE_KEY` or a keystore with `HL_KEYSTORE_PASSWORD` is required.
 
@@ -676,7 +683,7 @@ hl run my_strategies.my_strategy:MyStrategy -i ETH-PERP --tick 10
 
 ```bash
 pip install -e ".[dev]"
-pytest tests/ -v                  # 483 tests
+pytest tests/ -v                  # 1,317 tests
 ```
 
 ## Attribution 

diff --git a/cli/commands/mcp.py b/cli/commands/mcp.py
@@ -26,5 +26,5 @@ def mcp_serve(
         raise typer.Exit(1)
 
     server = create_mcp_server()
-    typer.echo(f"Starting MCP server (transport={transport}) ...")
+    typer.echo(f"Starting MCP server (transport={transport}) ...", err=True)
     server.run(transport=transport)
diff --git a/cli/commands/run.py b/cli/commands/run.py
@@ -53,7 +53,7 @@ def run_cmd(
     ),
     model: Optional[str] = typer.Option(
         None, "--model",
-        help="LLM model override for claude_agent strategy",
+        help="LLM model override for ai_agent strategy",
     ),
 ):
     """Start autonomous trading with a strategy."""

diff --git a/cli/commands/setup.py b/cli/commands/setup.py
@@ -70,11 +70,17 @@ def setup_check():
     else:
         ok_items.append("Builder fee: not configured (optional)")
 
-    # 5. LLM key (for claude_agent)
-    if os.environ.get("ANTHROPIC_API_KEY") or os.environ.get("GEMINI_API_KEY"):
-        ok_items.append("LLM API key found")
+    # 5. Hosted LLM key (only needed for the subscription-gated ai_agent product path)
+    if (
+        os.environ.get("OPENROUTER_API_KEY")
+        or os.environ.get("AI_API_KEY")
+        or os.environ.get("ANTHROPIC_API_KEY")
+        or os.environ.get("GEMINI_API_KEY")
+        or os.environ.get("OPENAI_API_KEY")
+    ):
+        ok_items.append("Hosted LLM API key found")
     else:
-        ok_items.append("LLM API key: not set (only needed for claude_agent strategy)")
+        ok_items.append("Hosted LLM API key: not set (only needed for ai_agent)")
 
     # 6. Data directories
     data_dir = Path("data/cli")

diff --git a/cli/mcp_server.py b/cli/mcp_server.py
@@ -2,6 +2,11 @@
 
 Fast tools (account, strategies, builder, wallet, setup) call Python directly.
 Long-running tools (run_strategy, apex_run, radar, reflect) use subprocess.
+
+Hosted MCP is the bring-your-own-agent path: a user's MCP-capable agent calls
+these tools, while web-auth handles wallet selection, consent, and scoped access.
+The OpenRouter-backed ai_agent strategy belongs to the separate hosted-agent
+product and is not required for generic MCP use.
 """
 from __future__ import annotations
 
@@ -38,10 +43,12 @@ def create_mcp_server():
     from mcp.server.fastmcp import FastMCP
 
     mcp = FastMCP(
-        "yex-trader",
+        "nunchi-trading",
         instructions=(
-            "Autonomous Hyperliquid trading CLI — 18 strategies, APEX orchestrator, "
-            "REFLECT reviews, and BTCSWP funding hedge proposals."
+            "Nunchi trading tools for MCP-capable agents. Bring your own agent; "
+            "web-auth selects the wallet and grants scoped tool access. Tools call "
+            "agent-cli for Hyperliquid/YEX trading, APEX orchestration, REFLECT "
+            "reviews, and BTCSWP funding hedge proposals."
         ),
     )
 
@@ -51,7 +58,12 @@ def create_mcp_server():
 
     @mcp.tool()
     def strategies() -> str:
-        """List all available trading strategies with descriptions and default parameters."""
+        """List available agent-cli strategies.
+
+        Generic MCP clients can use any deterministic strategy directly. The
+        ai_agent strategy is the Nunchi-hosted LLM agent product path and may
+        require hosted-agent subscription/provider configuration.
+        """
         from cli.strategy_registry import STRATEGY_REGISTRY, YEX_MARKETS
 
         result = {"strategies": {}, "yex_markets": {}}
@@ -124,7 +136,7 @@ def wallet_auto(save_env: bool = True) -> str:
 
     @mcp.tool()
     def setup_check() -> str:
-        """Validate environment — SDK, keys, network, builder fee."""
+        """Validate environment, web-auth pairing, network, and builder fee."""
         import os
         from cli.keystore import list_keystores
         from cli.config import TradingConfig
@@ -145,20 +157,21 @@ def setup_check() -> str:
         keystores = list_keystores()
         from cli.web_auth import get_stored_pairing
         pairing = get_stored_pairing()
-        if has_env_key:
+        if pairing is not None:
+            ok_items.append(f"Web-auth wallet selected ({pairing.selected_or_master_address})")
+        elif has_env_key:
             ok_items.append("HL_PRIVATE_KEY set")
-            if pairing is None:
-                warnings.append(
-                    "Raw-key mode active. Prefer hl pair connect or hosted Nunchi Auth for MCP/agent use."
-                )
+            warnings.append(
+                "Raw-key mode active. Prefer hl pair connect or hosted Nunchi Auth for MCP/agent use."
+            )
         elif keystores:
             ok_items.append(f"Keystore found ({len(keystores)} keys)")
         else:
             issues.append("No private key: set HL_PRIVATE_KEY or run wallet_auto")
-        if pairing is not None:
-            ok_items.append(f"Paired wallet active ({pairing.selected_or_master_address})")
-        else:
-            warnings.append("No paired wallet found. Run hl pair connect to enable browser-approved signing.")
+        if pairing is None:
+            warnings.append(
+                "No web-auth wallet selected. Run hl pair connect, or complete the hosted MCP connect flow."
+            )
 
         # Network
         testnet = os.environ.get("HL_TESTNET", "true").lower()
@@ -176,12 +189,16 @@ def setup_check() -> str:
             "ok": ok_items,
             "warnings": warnings,
             "issues": issues,
+            "web_auth": {
+                "paired": pairing is not None,
+                "selected_wallet": pairing.selected_or_master_address if pairing is not None else None,
+            },
             "passed": len(issues) == 0,
         }, indent=2)
 
     @mcp.tool()
     def pair_status() -> str:
-        """Show web-auth paired wallet status."""
+        """Show the web-auth paired wallet selected for MCP tool calls."""
         return _run_hl("pair", "status")
 
     @mcp.tool()

diff --git a/cli/skill.md b/cli/skill.md
@@ -94,8 +94,9 @@ hl apex run --mainnet                                   # APEX multi-slot
 | `HL_TESTNET` | No | `true` (default) or `false` for mainnet |
 | `BUILDER_ADDRESS` | No | Override builder fee address (default: hardcoded) |
 | `BUILDER_FEE_TENTHS_BPS` | No | Override fee rate (default: 100 = 10 bps) |
-| `ANTHROPIC_API_KEY` | No | For `claude_agent` strategy |
-| `GEMINI_API_KEY` | No | For `claude_agent` with Gemini |
+| `OPENROUTER_API_KEY` | No | For hosted `ai_agent` through OpenRouter |
+| `ANTHROPIC_API_KEY` | No | For `ai_agent` with Claude |
+| `GEMINI_API_KEY` | No | For `ai_agent` with Gemini |
 
 \* Either `HL_PRIVATE_KEY` or a keystore with `HL_KEYSTORE_PASSWORD` is required.
 
@@ -205,7 +206,7 @@ Tools: `strategies`, `builder_status`, `wallet_list`, `wallet_auto`, `setup_chec
 | aggressive_taker | Taker | Directional spread crossing with bias |
 | hedge_agent | Risk | Inventory exposure reducer; BTCSWP funding hedge lives under `hl hedge` |
 | rfq_agent | RFQ | Block-size dark RFQ liquidity |
-| claude_agent | LLM | Claude/Gemini-powered autonomous trading agent |
+| ai_agent | LLM | Hosted multi-provider autonomous trading agent |
 
 ## Instruments