Prompt Generation Guide
This guide helps you write effective prompts for the Monad MCP chat/agent system. Use it to get the best results for market analysis, agent code generation, contract development, and more.
1. General Tips
Be specific: State the asset, protocol, or task you want.
State your goal: Do you want analysis, code, a full agent, or a step-by-step guide?
Request code or contracts explicitly if needed ("Return Solidity contract or Return ts script").
For multi-step flows, list each step or requirement.
Ask for canonical addresses or rulebook lookups to avoid hallucinated data.
Use plain English—no need for special syntax.
2. Prompt Types
A. Market Analysis
"Analyze BTC. Give me a technical breakdown and a trading signal."
"Give me a high-alpha market regime analysis for AAPL."
B. Trading Agent Generation
"Generate a TypeScript trading bot for BTC that fetches price, places a GTC limit order, and cancels it."
"Deploy a margin-trading bot for ETH that deposits MON, mints aprMON, and trades on Kuru. Return both Solidity and TypeScript code."
C. Contract Development
"Write a Solidity contract that wraps MON → aprMON deposit, requestRedeem, and redeem, with events."
"Give me Hardhat deployment instructions for the above contract."
D. Multi-Protocol
Build a complete end-to-end project for Monad testnet that:
Deploys a Solidity helper contract to deposit MON into Kintsu and mint stMON.
Uses the stMON as collateral in Kuru’s MarginAccount.
Places and manages a GTC limit order on the MON/stMON market using the Kuru SDK."
Walk me through, step by step, how to go from a funded Monad testnet wallet to running a margin-trading bot that:
Deploys a Solidity helper contract for MON → aprMON deposit, requestRedeem, and redeem.
Approves the Kuru MarginAccount and Router contracts.
Uses a TypeScript script (ethers v6 + @kuru-labs/kuru-sdk) to: - Deposit MON through the helper. - Fetch market params with ParamFetcher.
3. Best Practices
If you want code, say so: "Return Solidity contract…" or "Return TypeScript script…"
For end-to-end flows, ask for both contract and script, and .env setup.
If you want explanations, add: "Explain each step" or "Add comments to all code."
For multi-asset or multi-protocol agents, list all assets/protocols you want included.
To avoid hallucinated addresses, ask: "Fetch canonical addresses from the rulebook."
4. Troubleshooting
If the response is too generic, add more detail to your prompt.
If you get hallucinated addresses, ask for a rulebook lookup.
For errors or missing code, clarify the output format you want (e.g., "Return only code blocks").
If you want a step-by-step guide, say so explicitly.
Last updated