> For the complete documentation index, see [llms.txt](https://asynchronus.gitbook.io/asynchronus-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://asynchronus.gitbook.io/asynchronus-docs/using-shell-user-guide.md). # Using Shell (User Guide) Welcome to Shell, your conversational gateway to the Asynchronus agent network! Shell allows you to command powerful AI agents to execute complex Web3 tasks using simple, natural language. Forget juggling multiple dApps, browser tabs, and block explorers – just tell Shell what you need. ### Getting Started with Shell Interacting with Shell is designed to be as intuitive as chatting with a knowledgeable assistant. 1. **Access Shell:** * Navigate to the Asynchronus web application (e.g., `https://app.asynchronus.ai` - *Note: Actual URL will be provided upon launch*). * You might also find Shell integrated into partner platforms or accessible via bots on platforms like Telegram or Discord in the future. 2. **Connect Your Wallet:** * To perform on-chain actions and analyze your specific holdings, you'll need to connect your Web3 wallet (like MetaMask, Rabby, Coinbase Wallet, etc.). * This connection allows Shell to identify your accounts (with your permission) and propose transactions for you to sign. **Asynchronus and Shell will NEVER have access to your private keys or be able to sign transactions without your explicit approval.** 3. **Your First Interaction:** * You'll see a conversational interface, likely a chat window. * Start by saying hello or asking about its capabilities! > **You:** Hi Shell, what can you do for me? > **Shell:** Hello! I am Shell, your autonomous conversational agent for Web3. I can help you with tasks like portfolio analysis, cross-chain bridging and staking, market research, token swaps, and more by coordinating specialized AI agents. What would you like to achieve today? ### How Shell Works When you give Shell a command, a sophisticated process unfolds behind the scenes: 1. **Natural Language Understanding:** Shell uses advanced AI models to understand your request, even if it's complex or involves multiple steps. It grasps your underlying *intent*. 2. **Task Decomposition:** Shell acts like an expert project manager. It breaks down your high-level goal into a series of smaller, actionable sub-tasks. For example, "Bridge ETH to Polygon and stake it" might become: * Find best bridge route (ETH Mainnet -> Polygon). * Check available ETH balance. * Find best staking pool on Polygon for ETH. * Check security/risk score of the pool. * Initiate bridge transaction. * Wait for bridge confirmation. * Initiate staking transaction. 3. **Sub-Agent Orchestration:** Shell assigns each sub-task to the most suitable **specialized AI agent** from the Asynchronus network (Analyze, Bridge, Swap, Security, etc.). 4. **Asynchronous Execution & Coordination:** These agents then work **in parallel** (asynchronously) whenever possible. They communicate through shared memory, passing results and coordinating their next steps. The Bridge Agent might work on finding a route while the Analyze Agent researches pools. 5. **User Interaction & Confirmation:** For actions involving fund movements or critical decisions, Shell will present the plan or specific transactions back to you for **review and approval** (via your connected wallet). 6. **Result Aggregation:** Once all sub-tasks are complete, Shell synthesizes the results and presents a clear summary or confirmation back to you. ### Example Workflows Let's see Shell in action with some common Web3 tasks: **Portfolio Analysis & Strategy** > **You:** Can you look at my 0x123...abc wallet on Arbitrum? Give me a breakdown of my assets and suggest two low-risk yield farming opportunities for my USDC. > **Shell:** *Working...* > > * *Analyze Agent: Fetching and classifying assets for 0x123...abc on Arbitrum.* > * *Analyze Agent: Scanning top Arbitrum DeFi protocols for USDC yield.* > * *Security Agent: Assessing risk profiles of identified pools (audits, TVL, age).* > > **Shell:** Okay, I see you hold mainly ETH, ARB, and USDC. For your USDC, here are two low-risk options: > > 1. **Aave v3:** Supply USDC for \~3.5% APY. Aave is a highly-audited, blue-chip lending protocol. > 2. **Curve (USDC/USDT Pool):** Provide liquidity for \~2.8% APY + Trading Fees. Curve is a stable-swap leader, known for low impermanent loss on stable pairs. Would you like me to guide you through depositing into one of these? **Cross-Chain Bridging & Staking** > **You:** I want to move 1000 GHO from Ethereum Mainnet to Avalanche and lend it on Aave there. Find the cheapest and fastest way. > **Shell:** *Working...* > > * *Bridge Agent: Comparing routes (e.g., Stargate, Router, Hop) for GHO (ETH -> AVAX) based on cost and time.* > * *Analyze Agent: Confirming Aave GHO lending market exists on Avalanche & checking APY.* > * *Security Agent: Verifying bridge and Aave contract addresses.* > > **Shell:** The optimal route appears to be via **Router Protocol**, estimated cost $8.50 and \~15 minutes. Aave on Avalanche offers \~4.1% APY for GHO. > > **Plan:** > > 1. Approve Router Protocol to spend 1000 GHO on Ethereum. > 2. Initiate the bridge transaction via Router. > 3. Once GHO arrives on Avalanche, approve Aave to spend it. > 4. Deposit GHO into Aave. > > Shall I prepare the first transaction for your approval? **Market Research** > **You:** What's the current developer activity and social sentiment for the Allora Network? > **Shell:** *Working...* > > * *Analyze Agent: Scanning GitHub repositories for Allora Network for recent commit frequency and contributor count.* > * *Analyze Agent: Querying social platforms (X/Twitter, Farcaster, Telegram) for mentions, sentiment scores, and key discussion points related to Allora.* > > **Shell:** Allora Network shows consistent high developer activity over the past 3 months, particularly in their core consensus and ML model integration modules. Social sentiment is generally positive (78/100), with excitement focused on their upcoming mainnet launch and recent partnerships. Some concerns noted around tokenomics clarity. Would you like a detailed report or links to key sources? **Advanced DeFi (Arbitrage, Rebalancing)** > **You:** Monitor WETH/USDC on Uniswap (Arbitrum) and KyberSwap (Arbitrum). If a price difference allows for at least 0.3% profit after gas fees on a $5000 trade, notify me immediately and prepare the swap sequence. > **Shell:** *Working...* > > * *Analyze Agent: Setting up real-time price feeds for WETH/USDC on Uniswap & KyberSwap (Arbitrum).* > * *Prediction Agent: Estimating potential gas costs for a 2-swap sequence.* > * *Swap Agent: Standing by to calculate routes and potential execution.* > > **Shell:** Understood. I am now monitoring WETH/USDC on Uniswap and KyberSwap on Arbitrum for a 0.3% arbitrage opportunity (net of estimated fees) on a $5000 trade size. I will notify you immediately if such an opportunity arises and present the proposed transactions. > > **⚠️ Important Disclaimer:** Arbitrage and automated trading carry significant risks, including price slippage, front-running, and smart contract vulnerabilities. While I can identify and prepare trades, **always review transactions carefully before signing**. Never risk more than you can afford to lose. Asynchronus is a tool; you are the ultimate decision-maker. *** This guide provides a starting point. The true power of Shell lies in its ability to understand and execute increasingly complex and personalized Web3 workflows. Don't hesitate to experiment and see what Asynchronus can do for you! --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://asynchronus.gitbook.io/asynchronus-docs/using-shell-user-guide.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.