cornerstone-autonomous-agent 2.0.0 → 2.1.0

package/README.md

@@ -1,12 +1,12 @@
-# Autonomous Agent – x402 MCP + LangChain.js
+# CornerStone MCP x402 Skill – Tools for Agents
 
 **Published as [cornerstone-autonomous-agent](https://www.npmjs.com/package/cornerstone-autonomous-agent)** on npm; **source: [FinTechTonic/autonomous-agent](https://github.com/FinTechTonic/autonomous-agent)**.
 
-Autonomous AI agent for **x402-paid MCP tools**: predict tickers, backtest strategies, link bank accounts, and query agent/borrower scores. Pays with **Aptos** or **EVM** via the x402 facilitator—no OpenAI key; uses **Hugging Face** for the LLM. Built for Cursor, OpenClaw/Moltbot, and headless runs.
+A **skill** that gives agents a set of **x402-paid MCP tools**: predict tickers, backtest strategies, link bank accounts, and query agent/borrower scores. Agents using this skill pay with **Aptos** or **EVM** via the x402 facilitator. Designed for marketplaces where agents can autonomously download and use the skill; also usable from Cursor, OpenClaw/Moltbot, and headless runners.
 
 ## Why?
 
-Most agent demos need a full backend and separate API keys. This agent talks to an **x402 MCP server**, pays with its own Aptos and EVM wallets (verify → settle), and uses an OpenAI-compatible LLM (e.g. Hugging Face). You get **run_prediction**, **run_backtest**, **link_bank_account**, and **score tools** (get_agent_reputation_score, get_borrower_score, by-email variants) with minimal setup—whitelist your agent, fund wallets, run. The server offers both Aptos and EVM payment options; the agent pays with whichever you prefer or have funded.
+Agents need tools, not a full backend. This skill provides **run_prediction**, **run_backtest**, **link_bank_account**, and **score tools** (get_agent_reputation_score, get_borrower_score, by-email variants). Agents using it talk to an **x402 MCP server**, pay with their own Aptos and EVM wallets (verify → settle), and get results after minimal setup—whitelist the agent’s addresses, fund wallets, then the agent can call the tools. The server offers both Aptos and EVM payment options; the skill handles 402 → pay → retry so the agent just calls the tool.
 
 ## Install
 
@@ -16,7 +16,7 @@ Most agent demos need a full backend and separate API keys. This agent talks to
 npm install cornerstone-autonomous-agent
 ```
 
-Copy the package’s `.env.example` into your project and set `MCP_SERVER_URL`, `X402_FACILITATOR_URL`, `HUGGINGFACE_API_KEY`, `LLM_MODEL`, and wallet paths. See [Config](#config).
+Copy the package’s `.env.example` into your project and set `X402_FACILITATOR_URL`, `HUGGINGFACE_API_KEY`, `LLM_MODEL`, and wallet paths. See [Config](#config).
 
 ### From source
 
@@ -25,8 +25,7 @@ git clone https://github.com/FinTechTonic/autonomous-agent.git && cd autonomous-
 npm install
 ```
 
-**OpenClaw / Moltbot:** See [adapters/openclaw/SKILL.md](adapters/openclaw/SKILL.md) or use your platform’s skill install (e.g. ClawHub). Then point `MCP_SERVER_URL` at your MCP server.
-
+**OpenClaw / Moltbot:** See [adapters/openclaw/SKILL.md](adapters/openclaw/SKILL.md) or use your platform’s skill install (e.g. ClawHub). 
 ## Quick Start
 
 **Important:** Another npm package is named [autonomous](https://www.npmjs.com/package/autonomous). To run **this** package without installing, always use the full name: **`npx cornerstone-autonomous-agent`**. After you `npm install cornerstone-autonomous-agent`, the short binary `autonomous` in that project runs this CLI.
@@ -62,7 +61,6 @@ Copy `.env.example` to `.env` and set:
 
 | Variable | Description |
 |----------|-------------|
-| `MCP_SERVER_URL` | x402 MCP server base URL (e.g. http://localhost:4023 or https://3001-borrower.replit.app) |
 | `X402_FACILITATOR_URL` | Facilitator base URL (Aptos + EVM verify/settle). Use public (e.g. https://x402-navy.vercel.app/facilitator) for full demo. |
 | `X402_EVM_FACILITATOR_URL` | Optional. EVM facilitator; defaults to X402_FACILITATOR_URL. |
 | `PREFERRED_PAYMENT_ORDER` | Optional. Comma-separated `network|asset` (e.g. `aptos:2|usdc,eip155:84532|usdc`) to choose payment option when server returns multiple. |
@@ -71,7 +69,7 @@ Copy `.env.example` to `.env` and set:
 | `LLM_MODEL` | Model ID (e.g. meta-llama/Llama-3.2-3B-Instruct) |
 | `APTOS_WALLET_PATH` | Aptos wallet JSON path. Multi-wallet: ~/.aptos-agent-wallets.json. |
 | `EVM_WALLET_PATH` | EVM wallet path. Multi-wallet: ~/.evm-wallets.json. Or set EVM_PRIVATE_KEY. |
-| `BASE_SEPOLIA_RPC` | Optional; Base Sepolia RPC. Agent supports all EVM chains in lib/chains.js (Base, Ethereum, Polygon, etc.). |
+| `BASE_SEPOLIA_RPC` | Optional; Base Sepolia RPC. Skill supports all EVM chains in lib/chains.js (Base, Ethereum, Polygon, etc.). |
 
 ## Commands
 
@@ -82,16 +80,16 @@ Use **`npx cornerstone-autonomous-agent <command>`** (or, in a project that has
 | `autonomous setup` | Generate EVM wallet (single; for multi use agent tool create_evm_wallet) |
 | `autonomous setup:aptos` | Generate Aptos wallet (single; for multi use create_aptos_wallet) |
 | `autonomous setup:evm:multichain` | Generate EVM multi-chain wallet |
-| `autonomous addresses` | Print all Aptos and EVM addresses for whitelisting at flow.html |
+| `autonomous addresses` | Print all Aptos and EVM addresses for whitelisting at https://arnstein.ch/flow.html |
 | `autonomous attest:aptos` | Sign attestation for Aptos wallet (set `APTOS_PRIVATE_KEY` or `APTOS_TESTNET_*`); submit to POST /attest/aptos |
 | `autonomous attest:evm` | Sign attestation for EVM wallet; submit to POST /attest/evm |
 | `autonomous balance [chain]` | EVM balance |
 | `autonomous transfer` | Transfer (see script for args) |
 | `autonomous contract` | Contract interaction helper |
 | `autonomous swap` | Swap helper |
-| `autonomous` or `autonomous agent` or `autonomous start [message]` | Run agent (default: demo balance + prediction) |
-| `npm run credit:aptos` | Credit Aptos agent (devnet: programmatic; testnet: instructions) |
-| `npx cornerstone-agent [message]` | Run agent (legacy bin name) |
+| `autonomous` or `autonomous agent` or `autonomous start [message]` | Run skill demo (default: balance + prediction) |
+| `npm run credit:aptos` | Credit Aptos wallet (devnet: programmatic; testnet: instructions) |
+| `npx cornerstone-agent [message]` | Run skill demo (legacy bin name) |
 
 **Crediting Aptos:** Testnet has no programmatic faucet—use [Aptos testnet faucet](https://aptos.dev/network/faucet). Devnet: `APTOS_FAUCET_NETWORK=devnet npm run credit:aptos`. See [Canteen – Aptos x402](https://canteenapp-aptos-x402.notion.site/).
 
@@ -102,11 +100,13 @@ Use **`npx cornerstone-autonomous-agent <command>`** (or, in a project that has
 | `run_prediction` | Stock prediction (symbol, horizon) | x402: Aptos or EVM |
 | `run_backtest` | Backtest trading strategy | x402: Aptos or EVM |
 | `link_bank_account` | CornerStone bank link (Plaid) | x402: Aptos or EVM |
-| `get_agent_reputation_score` | Agent reputation score (allowlisted wallet) | x402 or lender credits |
-| `get_borrower_score` | Borrower score (100 or 100+X when bank linked) | x402 or lender credits |
+| `get_agent_reputation_score` | Reputation Score: measures the agent's ability to transact using x402 | x402 or lender credits |
+| `get_borrower_score` | Borrower score: measures real borrower behavior | x402 or lender credits |
 | `get_agent_reputation_score_by_email` | Reputation score by email (resolves to agent) | x402 or lender credits |
 | `get_borrower_score_by_email` | Borrower score by email | x402 or lender credits |
 
+**Score definitions:** Reputation Score measures an agent’s ability to transact using x402. Borrower score measures real borrower behavior.
+
 All paid tools accept **both Aptos and EVM**; the server returns 402 with multiple options. Use `PREFERRED_PAYMENT_ORDER` to prefer one chain/asset.
 
 ## Supported Networks
@@ -119,11 +119,11 @@ All paid tools accept **both Aptos and EVM**; the server returns 402 with multip
 ## x402 Flow
 
 ```
-Agent calls MCP tool
+Agent (using this skill) calls MCP tool
   → Server returns 402 + payment requirements (single or array of options: USDC, APT, native ETH)
-  → Agent picks one option (by preferredPaymentOrder or first), builds payload (Aptos or EVM)
-  → Agent calls facilitator /verify then /settle
-  → Agent retries request with payment_payload
+  → Skill picks one option (by preferredPaymentOrder or first), builds payload (Aptos or EVM)
+  → Skill calls facilitator /verify then /settle
+  → Skill retries request with payment_payload
   → Server returns result + payment_receipt
 ```
 
@@ -141,7 +141,7 @@ autonomous/
 │   │   ├── evm/          # EVM wallet, signPayment (Base)
 │   │   └── x402/         # Payment types, verify/settle flow
 │   ├── cli.js            # CLI router (autonomous / cornerstone-autonomous-agent)
-│   ├── run-agent.js      # Agent entrypoint
+│   ├── run-agent.js      # Skill demo entrypoint (agent + tools)
 │   ├── setup.js          # EVM wallet generation
 │   ├── setup-aptos.js    # Aptos wallet generation
 │   ├── attest-aptos-wallet.js / attest-evm-wallet.js
@@ -151,26 +151,26 @@ autonomous/
 └── package.json
 ```
 
-**Core pieces:** `lib/mcp` — MCP client and 402 retry; `lib/aptos` / `lib/evm` — wallets and payment signing; `lib/x402` — verify/settle; `agent/` — LangChain.js ReAct agent and tools.
+**Core pieces:** `lib/mcp` — MCP client and 402 retry; `lib/aptos` / `lib/evm` — wallets and payment signing; `lib/x402` — verify/settle; `agent/` — LangChain.js ReAct runner and tools (for agents using the skill).
 
 ## Tech Stack
 
 - **Runtime:** Node.js 18+
-- **Agent:** LangChain.js (ReAct), OpenAI-compatible LLM (e.g. Hugging Face)
+- **Runner:** LangChain.js (ReAct), OpenAI-compatible LLM (e.g. Hugging Face) for the demo; agents use the tools via their own runtime
 - **MCP:** [Model Context Protocol](https://modelcontextprotocol.io) + x402 payment flow
 - **Chains:** Aptos (viem-style + @aptos-labs/ts-sdk), EVM (viem) for Base Sepolia/Base
 - **Payments:** x402 facilitator (verify/settle), local wallet storage
 
 ## Security
 
-- **Wallets:** Stored locally (e.g. `~/.aptos-agent-wallets.json`, `~/.evm-wallets.json`); private keys not logged or sent except as signed payloads to the facilitator.
+- **Wallets:** Stored locally (e.g. `~/.aptos-agent-wallets.json`, `~/.evm-wallets.json`) for the agent using the skill; private keys not logged or sent except as signed payloads to the facilitator.
 - **Payments:** Only verify/settle go to the facilitator; no custody of funds by the MCP server.
-- **Whitelist:** Agent addresses must be allowlisted at the onboarding flow before paid tools succeed.
+- **Whitelist:** Addresses used by the agent must be allowlisted at the onboarding flow before paid tools succeed.
 
 ## Capability + adapters
 
-- **Capability:** Core (`src/`) — MCP client, x402 flow, local tools. No OpenAI/Claw/Anthropic logic in code.
-- **Adapters:** `adapters/` — how each platform uses the capability:
+- **Capability:** Core (`src/`) — MCP client, x402 flow, tools for agents. No OpenAI/Claw/Anthropic logic in code.
+- **Adapters:** `adapters/` — how each platform or marketplace loads the skill:
   - [adapters/openclaw/SKILL.md](adapters/openclaw/SKILL.md) — OpenClaw / Moltbot
   - [adapters/openai/openapi.yaml](adapters/openai/openapi.yaml) — Custom GPTs / Assistants
   - [adapters/anthropic/tools.json](adapters/anthropic/tools.json) — Claude tools
@@ -179,8 +179,8 @@ autonomous/
 ## Deployment order
 
 1. **x402 facilitator** — Use public (e.g. https://x402-navy.vercel.app/facilitator) for full demo; or run local and set X402_EVM_FACILITATOR_URL to public for link_bank_account.
-2. **MCP server** — x402-enabled (run locally or use Replit MCP URL).
-3. **Agent** — `node src/run-agent.js` or PM2 (`pm2 start ecosystem.config.cjs --only agent-autonomous` from repo root).
+2. **MCP server** — x402-enabled.
+3. **Skill runner (demo)** — `node src/run-agent.js` or PM2 (`pm2 start ecosystem.config.cjs --only agent-autonomous` from repo root).
 
 ## References
 

package/adapters/local/README.md

@@ -1,10 +1,10 @@
 # Local / OSS agents
 
-Use the CreditNexus x402 capability from local or OSS agents (e.g. LM Studio, AutoGen, CrewAI).
+Use the CornerStone MCP x402 capability from local or OSS agents (e.g. LM Studio, AutoGen, CrewAI).
 
 1. **Source:** [FinTechTonic/autonomous-agent](https://github.com/FinTechTonic/autonomous-agent).
 2. **Install:** `npm install` in `autonomous/`.
-3. **Config:** Set `MCP_SERVER_URL` to your MCP server (e.g. `https://borrower.replit.app` or `http://localhost:4023`). Set x402 facilitator URL and LLM/env as needed.
+3. **Config:** Set x402 facilitator URL and LLM/env as needed.
 4. **Run:** `node src/run-agent.js "your message"` or wire MCP client to your agent framework.
 5. **x402:** On 402 responses, call facilitator verify → settle, then retry the request with `PAYMENT-SIGNATURE` header.
 

package/adapters/openai/openapi.yaml

@@ -1,11 +1,11 @@
 openapi: 3.0.3
 info:
-  title: CreditNexus x402 MCP Tools
+  title: CornerStone MCP x402 Tools
   description: Payment-protected tools (run_prediction, run_backtest, open_bank_account, scores). Use x402 flow (402 → pay → retry with PAYMENT-SIGNATURE).
   version: 1.0.0
 servers:
-  - url: http://localhost:4023
-    description: MCP server (override with MCP_SERVER_URL)
+  - url: https://arnstein.ch/mcp
+    description: MCP server
 paths:
   /mcp/prediction/{symbol}:
     get:

package/adapters/openclaw/SKILL.md

@@ -1,62 +1,73 @@
 ---
 name: autonomous-agent
-description: CreditNexus x402 agent. Use when the user wants stock predictions, backtests, bank linking, or agent/borrower scores. Payment-protected MCP tools (run_prediction, run_backtest, link_bank_account, get_agent_reputation_score, get_borrower_score, and by-email variants) with x402 flow (Aptos + Base). Agent handles 402 → pay → retry autonomously. Supports wallet attestation (signing) for onboarding.
+description: CornerStone MCP x402 skill for agents. Tools for stock predictions, backtests, bank linking, and agent/borrower scores. Payment-protected MCP tools with x402 flow (Aptos + Base). Skill handles 402 → pay → retry. Wallet attestation for onboarding. For marketplaces where agents download and use skills autonomously.
 metadata: {"openclaw":{"emoji":"📈","homepage":"https://github.com/FinTechTonic/autonomous-agent","requires":{"bins":["node","npm"]},"primaryEnv":"MCP_SERVER_URL","skillKey":"autonomous-agent"},"clawdbot":{"emoji":"📈","homepage":"https://github.com/FinTechTonic/autonomous-agent","requires":{"bins":["node","npm"]}}}
 ---
 
-# CreditNexus x402 Agent Skill
+# CornerStone MCP x402 Skill (for Agents)
 
-Autonomous agent that calls x402-protected MCP tools: stock prediction, backtest, bank linking, and agent/borrower scores. Handles payment flow (402 → pay → retry with `payment_payload`) with Aptos (prediction/backtest) and Base (banking). Supports **wallet attestation** (signing) for onboarding (POST /attest/aptos, /attest/evm).
+Skill that gives **agents** tools to call x402-protected MCP endpoints: stock prediction, backtest, bank linking, and agent/borrower scores. **Payment is automatic** — the skill handles 402 → sign → verify → settle → retry transparently. Supports **wallet attestation** (signing) for onboarding (POST /attest/aptos, /attest/evm).
 
 ## Installation
 
 Clone or copy the repo. When loaded from OpenClaw/MoltBook, the skill folder is `{baseDir}`; run commands from the **repo root** (parent of `adapters/openclaw` or of `skills/autonomous-agent`).
 
 ```bash
-# From repository root
 git clone https://github.com/FinTechTonic/autonomous-agent.git && cd autonomous-agent
 npm install
 ```
 
-Set `MCP_SERVER_URL` to your MCP server (e.g. `https://borrower.replit.app`). Copy `.env.example` to `.env` and set:
+Copy `.env.example` to `.env` and set:
 
-- `MCP_SERVER_URL` – MCP server base URL (MCP protocol at `/mcp`)
 - `X402_FACILITATOR_URL` – x402 facilitator (verify/settle)
 - `LLM_BASE_URL`, `HUGGINGFACE_API_KEY` or `HF_TOKEN`, `LLM_MODEL` – for inference
 - `APTOS_WALLET_PATH`, `EVM_WALLET_PATH` (or `EVM_PRIVATE_KEY`) – for payments
 
-## Run the agent
+## Quick-start workflow
 
-From the **repository root** (where `package.json` and `src/` live):
+1. `get_wallet_addresses()` – check what wallets exist.
+2. If empty: `create_aptos_wallet()` + `create_evm_wallet()`.
+3. Fund: `credit_aptos_wallet()` + `fund_evm_wallet()`.
+4. Whitelist addresses at https://arnstein.ch/flow.html.
+5. Check balances: `balance_aptos()`, `balance_evm({ chain: "baseSepolia" })`.
+6. Call paid tools: `run_prediction`, `run_backtest`, `link_bank_account`, or score tools.
+
+## Run the skill (demo)
 
 ```bash
 npx cornerstone-agent "Run a 30-day prediction for AAPL"
-# Or interactive
 npx cornerstone-agent
-# Or from repo: npm run agent -- "..." or node src/run-agent.js "..."
+npm run agent -- "..."
+node src/run-agent.js "..."
 ```
 
-**x402 flow:** Agent calls tool without `payment_payload` → server returns 402 + `paymentRequirements` → agent signs, facilitator verify/settle → agent retries with `payment_payload` → receives result + `paymentReceipt`.
-
 ## Wallet attestation (signing)
 
-To prove wallet ownership during onboarding, run from repo root:
-
 - Aptos: `npm run attest:aptos` or `npx cornerstone-agent-attest-aptos` — output to POST /attest/aptos
 - EVM: `npm run attest:evm` or `npx cornerstone-agent-attest-evm` — output to POST /attest/evm
 
-## MCP Tools
-
-All tools are on the MCP server at `/mcp`. See repo [MCP_INTEGRATION_REFERENCE.md](https://github.com/FinTechTonic/autonomous-agent/blob/main/MCP_INTEGRATION_REFERENCE.md) for resources and pricing.
-
-| Tool | Resource | Description | Cost |
-|------|----------|-------------|------|
-| `run_prediction` | `/mcp/prediction/{symbol}` | Stock prediction (symbol, horizon) | ~6¢ |
-| `run_backtest` | `/mcp/backtest/{symbol}` | Backtest (symbol, start/end, strategy) | ~6¢ |
-| `link_bank_account` | `/mcp/banking/link` | CornerStone/Plaid bank link token | ~5¢ (config) |
-| `get_agent_reputation_score` | `/mcp/scores/reputation` | Agent reputation (100 allowlisted); x402 or lender credits | ~6¢ |
-| `get_borrower_score` | `/mcp/scores/borrower` | Borrower score (100 or 100+Plaid); x402 or lender credits | ~6¢ |
-| `get_agent_reputation_score_by_email` | `/mcp/scores/reputation-by-email` | Reputation by email; requires SCORE_BY_EMAIL_ENABLED | base + extra |
-| `get_borrower_score_by_email` | `/mcp/scores/borrower-by-email` | Borrower score by email; requires SCORE_BY_EMAIL_ENABLED | base + extra |
-
-Whitelist your agent at the onboarding flow (e.g. `MCP_SERVER_URL/flow.html`) so the server allows your wallet.
+## Tool reference
+
+### Wallet tools (local)
+| Tool | Args | Returns |
+|------|------|---------|
+| `get_wallet_addresses` | none | `{ aptos: [{ address, network }], evm: [...] }` |
+| `create_aptos_wallet` | `{ force?, network? }` | `{ success, address, network }` |
+| `create_evm_wallet` | `{ force?, network? }` | `{ success, address, network }` |
+| `credit_aptos_wallet` | `{ amount_octas? }` | devnet: funds directly; testnet: `{ faucet_url, address }` |
+| `fund_evm_wallet` | none | `{ faucet_url, address, message }` |
+| `balance_aptos` | none | `{ address, balances: { usdc, apt } }` |
+| `balance_evm` | `{ chain? }` | `{ address, chain, balance, symbol }` |
+
+### Paid MCP tools (x402 — payment automatic)
+| Tool | Args | Returns | Cost |
+|------|------|---------|------|
+| `run_prediction` | `{ symbol, horizon? }` | Forecast data | ~6¢ |
+| `run_backtest` | `{ symbol, startDate?, endDate?, strategy? }` | Performance metrics | ~6¢ |
+| `link_bank_account` | none | `{ link_token }` | ~5¢ |
+| `get_agent_reputation_score` | `{ agent_address?, payer_wallet? }` | `{ reputation_score }` | ~6¢ or credits |
+| `get_borrower_score` | `{ agent_address?, payer_wallet? }` | `{ score }` | ~6¢ or credits |
+| `get_agent_reputation_score_by_email` | `{ email, payer_wallet? }` | `{ reputation_score }` | higher |
+| `get_borrower_score_by_email` | `{ email, payer_wallet? }` | `{ score }` | higher |
+
+Whitelist the addresses the agent uses at https://arnstein.ch/flow.html so the server allows those wallets.

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "cornerstone-autonomous-agent",
-  "version": "2.0.0",
-  "description": "CornerStone Agentic Score: autonomous agent that uses an x402-enabled MCP server to predict tickers, backtest trading strategies, and open bank accounts. Handles 402 Payment Required (pay via Aptos/Ethereum facilitator and retry). LangChain.js ReAct agent with Hugging Face inference (OpenAI-compatible endpoint).",
+  "version": "2.1.0",
+  "description": "CornerStone MCP x402 skill for agents: tools to predict tickers, backtest strategies, link bank accounts, and query agent/borrower scores. Agents pay via Aptos/EVM (x402). Handles 402 → pay → retry. Publishable in marketplaces for autonomous download and use. Includes LangChain.js demo runner (Hugging Face / OpenAI-compatible).",
   "type": "module",
   "main": "src/run-agent.js",
   "bin": {
@@ -51,7 +51,7 @@
     "@aptos-labs/ts-sdk": "^1.0.0"
   },
   "keywords": [
-    "creditnexus",
+    "cornerstone-mcp",
     "demo",
     "autonomous-agent",
     "x402",
@@ -67,7 +67,7 @@
     "evm",
     "crypto"
   ],
-  "author": "CreditNexus",
+  "author": "CornerStone MCP",
   "license": "GPL-2.0-only",
   "repository": {
     "type": "git",

package/skills/README.md

@@ -4,6 +4,6 @@ This folder follows the **AgentSkills** layout expected by MoltBook and OpenClaw
 
 | Skill | Description |
 |-------|-------------|
-| [autonomous-agent](autonomous-agent/SKILL.md) | x402 MCP agent: stock prediction, backtest, link_bank_account, agent/borrower scores (Aptos + Base). Wallet attestation (signing) for onboarding. |
+| [autonomous-agent](autonomous-agent/SKILL.md) | x402 MCP skill for agents: stock prediction, backtest, link_bank_account, agent/borrower scores (Aptos + Base). Wallet attestation for onboarding. |
 
 **Loading:** Add this repo path to `skills.load.extraDirs` in `~/.openclaw/openclaw.json` (MoltBook/OpenClaw will load `autonomous-agent` from `skills/autonomous-agent/`), or copy `autonomous-agent` into `~/.openclaw/skills/`, or use the repo as your OpenClaw workspace. See the main [README](../README.md#moltbook--openclaw).

package/skills/autonomous-agent/SKILL.md

@@ -1,62 +1,73 @@
 ---
 name: autonomous-agent
-description: CreditNexus x402 agent. Use when the user wants stock predictions, backtests, bank linking, or agent/borrower scores. Payment-protected MCP tools (run_prediction, run_backtest, link_bank_account, get_agent_reputation_score, get_borrower_score, and by-email variants) with x402 flow (Aptos + Base). Agent handles 402 → pay → retry autonomously. Supports wallet attestation (signing) for onboarding.
+description: CornerStone MCP x402 skill for agents. Tools for stock predictions, backtests, bank linking, and agent/borrower scores. Payment-protected MCP tools with x402 flow (Aptos + Base). Skill handles 402 → pay → retry. Wallet attestation for onboarding. For marketplaces where agents download and use skills autonomously.
 metadata: {"openclaw":{"emoji":"📈","homepage":"https://github.com/FinTechTonic/autonomous-agent","requires":{"bins":["node","npm"]},"primaryEnv":"MCP_SERVER_URL","skillKey":"autonomous-agent"},"clawdbot":{"emoji":"📈","homepage":"https://github.com/FinTechTonic/autonomous-agent","requires":{"bins":["node","npm"]}}}
 ---
 
-# CreditNexus x402 Agent Skill
+# CornerStone MCP x402 Skill (for Agents)
 
-Autonomous agent that calls x402-protected MCP tools: stock prediction, backtest, bank linking, and agent/borrower scores. Handles payment flow (402 → pay → retry with `payment_payload`) with Aptos (prediction/backtest) and Base (banking). Supports **wallet attestation** (signing) for onboarding (POST /attest/aptos, /attest/evm).
+Skill that gives **agents** tools to call x402-protected MCP endpoints: stock prediction, backtest, bank linking, and agent/borrower scores. **Payment is automatic** — the skill handles 402 → sign → verify → settle → retry transparently. Supports **wallet attestation** (signing) for onboarding (POST /attest/aptos, /attest/evm).
 
 ## Installation
 
 When this skill is loaded from the **autonomous-agent** repo, the repo root is the parent of the skill folder `{baseDir}`. Clone and install from the repo root:
 
 ```bash
-# From repository root (parent of {baseDir} when using this repo)
 git clone https://github.com/FinTechTonic/autonomous-agent.git && cd autonomous-agent
 npm install
 ```
 
-Set `MCP_SERVER_URL` to your MCP server (e.g. `https://borrower.replit.app`). Copy `.env.example` to `.env` and set:
+Copy `.env.example` to `.env` and set:
 
-- `MCP_SERVER_URL` – MCP server base URL (MCP protocol at `/mcp`)
-- `X402_FACILITATOR_URL` – x402 facilitator (verify/settle)
+- set `X402_FACILITATOR_URL` too
 - `LLM_BASE_URL`, `HUGGINGFACE_API_KEY` or `HF_TOKEN`, `LLM_MODEL` – for inference
 - `APTOS_WALLET_PATH`, `EVM_WALLET_PATH` (or `EVM_PRIVATE_KEY`) – for payments
 
-## Run the agent
+## Quick-start workflow
 
-From the **repository root** (where `package.json` and `src/` live):
+1. `get_wallet_addresses()` – check what wallets exist.
+2. If empty: `create_aptos_wallet()` + `create_evm_wallet()`.
+3. Fund: `credit_aptos_wallet()` + `fund_evm_wallet()`.
+4. Whitelist addresses at https://arnstein.ch/sse/flow.html.
+5. Check balances: `balance_aptos()`, `balance_evm({ chain: "baseSepolia" })`.
+6. Call paid tools: `run_prediction`, `run_backtest`, `link_bank_account`, or score tools.
+
+## Run the skill (demo)
 
 ```bash
 npx cornerstone-agent "Run a 30-day prediction for AAPL"
-# Or interactive
 npx cornerstone-agent
-# Or from repo: npm run agent -- "..." or node src/run-agent.js "..."
+npm run agent -- "..."
+node src/run-agent.js "..."
 ```
 
-**x402 flow:** Agent calls tool without `payment_payload` → server returns 402 + `paymentRequirements` → agent signs, facilitator verify/settle → agent retries with `payment_payload` → receives result + `paymentReceipt`.
-
 ## Wallet attestation (signing)
 
-To prove wallet ownership during onboarding, run from repo root:
-
 - Aptos: `npm run attest:aptos` or `npx cornerstone-agent-attest-aptos` — output to POST /attest/aptos
 - EVM: `npm run attest:evm` or `npx cornerstone-agent-attest-evm` — output to POST /attest/evm
 
-## MCP Tools
-
-All tools are on the MCP server at `/mcp`. See repo [MCP_INTEGRATION_REFERENCE.md](https://github.com/FinTechTonic/autonomous-agent/blob/main/MCP_INTEGRATION_REFERENCE.md) for resources and pricing.
-
-| Tool | Resource | Description | Cost |
-|------|----------|-------------|------|
-| `run_prediction` | `/mcp/prediction/{symbol}` | Stock prediction (symbol, horizon) | ~6¢ |
-| `run_backtest` | `/mcp/backtest/{symbol}` | Backtest (symbol, start/end, strategy) | ~6¢ |
-| `link_bank_account` | `/mcp/banking/link` | CornerStone/Plaid bank link token | ~5¢ (config) |
-| `get_agent_reputation_score` | `/mcp/scores/reputation` | Agent reputation (100 allowlisted); x402 or lender credits | ~6¢ |
-| `get_borrower_score` | `/mcp/scores/borrower` | Borrower score (100 or 100+Plaid); x402 or lender credits | ~6¢ |
-| `get_agent_reputation_score_by_email` | `/mcp/scores/reputation-by-email` | Reputation by email; requires SCORE_BY_EMAIL_ENABLED | base + extra |
-| `get_borrower_score_by_email` | `/mcp/scores/borrower-by-email` | Borrower score by email; requires SCORE_BY_EMAIL_ENABLED | base + extra |
-
-Whitelist your agent at the onboarding flow (e.g. `MCP_SERVER_URL/flow.html`) so the server allows your wallet.
+## Tool reference
+
+### Wallet tools (local)
+| Tool | Args | Returns |
+|------|------|---------|
+| `get_wallet_addresses` | none | `{ aptos: [{ address, network }], evm: [...] }` |
+| `create_aptos_wallet` | `{ force?, network? }` | `{ success, address, network }` |
+| `create_evm_wallet` | `{ force?, network? }` | `{ success, address, network }` |
+| `credit_aptos_wallet` | `{ amount_octas? }` | devnet: funds directly; testnet: `{ faucet_url, address }` |
+| `fund_evm_wallet` | none | `{ faucet_url, address, message }` |
+| `balance_aptos` | none | `{ address, balances: { usdc, apt } }` |
+| `balance_evm` | `{ chain? }` | `{ address, chain, balance, symbol }` |
+
+### Paid MCP tools (x402 — payment automatic)
+| Tool | Args | Returns | Cost |
+|------|------|---------|------|
+| `run_prediction` | `{ symbol, horizon? }` | Forecast data | ~6¢ |
+| `run_backtest` | `{ symbol, startDate?, endDate?, strategy? }` | Performance metrics | ~6¢ |
+| `link_bank_account` | none | `{ link_token }` | ~5¢ |
+| `get_agent_reputation_score` | `{ agent_address?, payer_wallet? }` | `{ reputation_score }` | ~6¢ or credits |
+| `get_borrower_score` | `{ agent_address?, payer_wallet? }` | `{ score }` | ~6¢ or credits |
+| `get_agent_reputation_score_by_email` | `{ email, payer_wallet? }` | `{ reputation_score }` | higher |
+| `get_borrower_score_by_email` | `{ email, payer_wallet? }` | `{ score }` | higher |
+
+Whitelist the addresses the agent uses at https://arnstein.ch/sse/flow.html so the server allows those wallets.

package/src/agent/agent.js

@@ -1,53 +1,77 @@
-/**
- * LangChain.js ReAct agent: MCP tools + local tools, Hugging Face LLM.
- */
-
-import { createReactAgent } from '@langchain/langgraph/prebuilt';
-import { createLLM } from './llm.js';
-import { createMcpTools } from './tools/mcpTools.js';
-import { createLocalTools } from './tools/localTools.js';
-
-const SYSTEM_MESSAGE = `You are an autonomous agent that can create, fund, and use your own Aptos and EVM wallets (optionally testnet or mainnet), then run stock predictions, backtests, and link bank accounts via paid MCP tools.
-
-Before trying any paid or wallet-dependent action: use get_wallet_addresses to see which wallets are configured. It returns lists: aptos: [{ address, network? }, ...], evm: [{ address, network? }, ...]. You can have multiple Aptos and multiple EVM wallets (e.g. one testnet and one mainnet each). If none exist, use create_aptos_wallet and create_evm_wallet (optionally pass network: "testnet" or "mainnet"; default is testnet). Use force: true to add another wallet when one already exists. Then tell the user to whitelist all addresses at http://localhost:4024/flow.html — the whitelist form supports multiple EVM and multiple Aptos rows with an optional testnet/mainnet tag per address. Use credit_aptos_wallet to fund the Aptos wallet (devnet programmatic or testnet instructions). Use fund_evm_wallet to get funding instructions for the EVM wallet.
-
-When the user asks to create or produce wallets: use create_aptos_wallet and create_evm_wallet (with network if they want testnet or mainnet), then credit_aptos_wallet and fund_evm_wallet for instructions, and remind the user to whitelist every address at http://localhost:4024/flow.html (add multiple EVM and Aptos rows as needed; tag testnet/mainnet if relevant).
-
-Use run_prediction for stock predictions (symbol, horizon in days). Use run_backtest for backtesting a strategy.
-Use link_bank_account to start the bank linking flow.
-Use get_agent_reputation_score and get_borrower_score to query scores for allowlisted wallets; use get_agent_reputation_score_by_email and get_borrower_score_by_email when you have an email (requires SCORE_BY_EMAIL_ENABLED on server). Score tools can be paid via x402 or lender credits (payer_wallet).
-Use balance_aptos and balance_evm to check wallet balances.
-When you need to pay for a tool (402), the payment is handled automatically; just call the tool.`;
-
-/**
- * Create agent graph: llm + tools (MCP + local).
- * @param {{ llm?: import('@langchain/core/language_models/chat_models').BaseChatModel; tools?: import('@langchain/core/tools').StructuredToolInterface[] }} options - llm and tools (if omitted, created from env and mcpClient)
- * @param {{ callTool: (name: string, args: Object) => Promise<Object> }} [options.mcpClient] - required if tools not provided
- * @returns {Promise<{ agent: import('@langchain/langgraph').CompiledStateGraph; runAgent: (message: string) => Promise<Object> }>}
- */
-export async function createAgent(options = {}) {
-  const llm = options.llm || createLLM();
-  let tools = options.tools;
-  if (!tools && options.mcpClient) {
-    tools = [...createMcpTools(options.mcpClient), ...createLocalTools()];
-  }
-  if (!tools) {
-    throw new Error('Provide options.tools or options.mcpClient to createAgent.');
-  }
-
-  const agent = createReactAgent({
-    llm,
-    tools,
-    stateModifier: (state) => [{ role: 'system', content: SYSTEM_MESSAGE }, ...(state.messages || [])],
-  });
-
-  async function runAgent(userMessage) {
-    const result = await agent.invoke(
-      { messages: [{ role: 'user', content: userMessage }] },
-      { recursionLimit: 50 }
-    );
-    return result;
-  }
-
-  return { agent, runAgent };
-}
+/**
+ * LangChain.js ReAct runner for the skill: MCP tools + local tools, Hugging Face LLM.
+ */
+
+import { createReactAgent } from '@langchain/langgraph/prebuilt';
+import { createLLM } from './llm.js';
+import { createMcpTools } from './tools/mcpTools.js';
+import { createLocalTools } from './tools/localTools.js';
+
+const SYSTEM_MESSAGE = `You are an agent with a skill that provides tools for wallet management and x402-paid MCP tools (predictions, backtests, bank linking, scores). Payment is automatic — just call the tool.
+
+## ALWAYS DO FIRST
+Before any paid or wallet action, call get_wallet_addresses (no args). It returns:
+  { aptos: [{ address, network }], evm: [{ address, network }] }
+- If aptos is empty and you need Aptos tools: create_aptos_wallet -> credit_aptos_wallet -> tell user to whitelist at https://arnstein.ch/flow.html
+- If evm is empty and you need EVM tools: create_evm_wallet -> fund_evm_wallet -> tell user to whitelist at https://arnstein.ch/flow.html
+- If wallets exist: check balance before paid calls. balance_aptos (for prediction/backtest/scores) or balance_evm({ chain: "baseSepolia" }) (for link_bank_account).
+
+## TOOLS
+
+Wallet tools:
+- get_wallet_addresses() -> { aptos: [...], evm: [...] }
+- create_aptos_wallet({ force?, network? }) -> { success, address, network }
+- create_evm_wallet({ force?, network? }) -> { success, address, network }
+- credit_aptos_wallet({ amount_octas? }) -> devnet: funds directly; testnet: returns faucet_url + instructions
+- fund_evm_wallet() -> returns faucet_url + address (manual funding)
+- balance_aptos() -> { address, balances: { usdc, apt } }
+- balance_evm({ chain? }) -> { address, chain, balance, symbol }. Chains: base, baseSepolia, ethereum, polygon, arbitrum, optimism
+
+Paid MCP tools (payment handled for you — just call them):
+- run_prediction({ symbol, horizon? }) — stock prediction, ~6c USDC. Example: run_prediction({ symbol: "AAPL", horizon: 30 })
+- run_backtest({ symbol, startDate?, endDate?, strategy? }) — backtest, ~6c. Dates "YYYY-MM-DD", strategy default "chronos"
+- link_bank_account() — Plaid bank link, ~5c on Base. Needs funded EVM wallet (baseSepolia for testnet)
+- get_agent_reputation_score({ agent_address?, payer_wallet? }) — reputation score
+- get_borrower_score({ agent_address?, payer_wallet? }) — borrower score
+- get_agent_reputation_score_by_email({ email, payer_wallet? }) — by email (needs SCORE_BY_EMAIL_ENABLED)
+- get_borrower_score_by_email({ email, payer_wallet? }) — by email
+
+## ERROR RECOVERY
+- "No Aptos/EVM wallet" -> create the missing wallet
+- "already exist" -> use existing wallet, or pass force: true to add another
+- "Payment verification failed" -> insufficient funds; check balance, tell user to fund
+- "403" / "not allowlisted" -> tell user to whitelist at https://arnstein.ch/flow.html
+- Timeout -> retry once`;
+
+/**
+ * Create agent graph: llm + tools (MCP + local).
+ * @param {{ llm?: import('@langchain/core/language_models/chat_models').BaseChatModel; tools?: import('@langchain/core/tools').StructuredToolInterface[] }} options - llm and tools (if omitted, created from env and mcpClient)
+ * @param {{ callTool: (name: string, args: Object) => Promise<Object> }} [options.mcpClient] - required if tools not provided
+ * @returns {Promise<{ agent: import('@langchain/langgraph').CompiledStateGraph; runAgent: (message: string) => Promise<Object> }>}
+ */
+export async function createAgent(options = {}) {
+  const llm = options.llm || createLLM();
+  let tools = options.tools;
+  if (!tools && options.mcpClient) {
+    tools = [...createMcpTools(options.mcpClient), ...createLocalTools()];
+  }
+  if (!tools) {
+    throw new Error('Provide options.tools or options.mcpClient to createAgent.');
+  }
+
+  const agent = createReactAgent({
+    llm,
+    tools,
+    stateModifier: (state) => [{ role: 'system', content: SYSTEM_MESSAGE }, ...(state.messages || [])],
+  });
+
+  async function runAgent(userMessage) {
+    const result = await agent.invoke(
+      { messages: [{ role: 'user', content: userMessage }] },
+      { recursionLimit: 50 }
+    );
+    return result;
+  }
+
+  return { agent, runAgent };
+}