Crypto Agent Design Prompt
Specifies an AI crypto agent's tools, permissions, guardrails and evaluation before a single line is written.
/ quick answer
Use to design a research or monitoring agent, and to decide whether it should ever be allowed to sign. Specifies an AI crypto agent's tools, permissions, guardrails and evaluation before a single line is written.
Design an AI agent for this crypto task. Security boundaries are part of the specification. TASK: [what the agent should do] Data it needs: [LIST] Should it execute transactions? [YES / NO] Available tools: [model, RPC/indexer, analytics API, wallet type, automation platform] Produce: 1. Scope: exactly what the agent does and what it explicitly does not do. 2. Tool list with typed inputs and outputs, split into read tools and write tools. 3. System prompt: role, method, output format, and refusal conditions. 4. Data validation: how the agent verifies inputs before reasoning on them. 5. Guardrails in deterministic code, not in the prompt: allowlists, value caps, rate limits, simulation, kill switch. 6. Human-in-the-loop points and what is shown to the human for approval. 7. Failure modes specific to crypto: hallucinated addresses, stale prices, spoofed tokens, MEV exposure. 8. Evaluation set: 10 test cases including adversarial inputs, with expected behaviour. 9. Logging and audit requirements. If the task does not require signing, say so and design a read-only agent.
Scope: monitor 15 wallets, research flagged tokens, produce a briefing. Does NOT sign. Read tools: get_wallet_activity, get_token_liquidity, get_contract_facts. Write tools: none. Refusals: no price predictions, no advice, no unverifiable claims. Validation: every address checksummed and resolved against the indexer before use. Guardrails: rate limit, 30-min data staleness cap. Eval: includes a spoofed token with a lookalike symbol; expected behaviour is flagging the mismatch.
/ frequently asked
What does the Crypto Agent Design Prompt prompt do?
Use to design a research or monitoring agent, and to decide whether it should ever be allowed to sign.
Which AI models work with this prompt?
It is model-agnostic: it works with any capable general model. Replace the bracketed variables with your own context before running it.
How do I use this prompt?
1) Copy the prompt with the copy button. 2) Replace every bracketed variable with your real details. 3) Paste it into your AI assistant. 4) Compare the answer with the example output and ask for changes if needed.
What output should I expect?
Scope: monitor 15 wallets, research flagged tokens, produce a briefing. Does NOT sign. Read tools: get_wallet_activity, get_token_liquidity, get_contract_facts. Write tools: none. Refusals: no price predictions, no advice, no unverifiable claims. Validation: every address checksummed and resolved ag.
/ continue exploring
Related concepts
The vocabulary this page depends on.
- →On-chain Agent
An on-chain agent is an AI agent with read access to blockchain data and a tightly scoped ability to propose or execute transactions.
- →Crypto Automation
Crypto automation is rule-based execution of monitoring, alerting and recurring on-chain actions, so decisions are made once and applied consistently.
- →Browser Agent
A browser agent drives a real web browser — navigating, clicking, typing and reading pages — to complete tasks on sites that have no API.
- →Web3 Wallet
A Web3 wallet stores your keys and signs transactions, letting you hold assets and connect to on-chain applications without an intermediary.
Related workflows
Turn this into a repeatable process.
- →AI Agent + Web3 Execution
Give an AI agent on-chain capability safely: read freely, propose openly, and sign only inside hard-coded limits.
- →Automate Agent Task Delegation
This workflow automates the process of an initiating AI agent delegating a complex task to a specialized recipient AI agent using a standardized Agent-to-Agent (A2A) protocol, ensuring efficient task handoff and execution. It streamlines multi-agent collaboration for defined objectives.
- →Automated Crypto Alerts
Define the conditions that would change a decision, monitor them automatically, and receive one clean notification instead of watching charts.
- →Automate Repetitive Web3 Actions
Replace manual claims, approvals and rebalances with scoped automations that cannot spend more than you allowed.
Related tool stacks
The tools that run it in production.
- →AI Agent + Web3 Stack
Agent framework, MCP/API tools, blockchain data access and a limited signing layer — with policy enforced in code.
- →Crypto Automation Stack
Automation platform, data APIs, alerting and optional wallet execution — the operational layer for monitoring and recurring actions.
- →AI Crypto Research Stack
A read-only research stack combining an AI assistant, web search, market data and on-chain analytics to screen assets quickly.
Related prompts
Reusable prompts for this job.
- →Recurring Crypto Workflow Prompt
Specifies a safe recurring on-chain automation with spend caps, idempotency, failure handling and a kill switch.
- →Crypto Research Agent Prompt
System prompt for a research agent that must cite sources, separate fact from inference, and refuse to predict prices.
- →Agent Architecture Spec Prompt
Turns a fuzzy agent idea into a reviewable five-layer architecture spec.
- →Multi-Agent Role Definition Prompt
Generates crisp role prompts and handoff contracts for a team of agents.
Related use cases
How people apply it, and what came out.
- →Build An AI Crypto Research Agent
A read-only research agent produced daily briefings on a 30-token watchlist, cutting a 90-minute manual routine to a 10-minute review.
- →Monitor A Crypto Portfolio
Aggregating six addresses across three chains revealed that a portfolio believed to hold 14 positions actually held one concentrated bet.
- →Automate Repetitive Web3 Actions
Automating reward claims and gas top-ups with session keys and spend caps saved about 5 hours per month with no unbounded authority.
Comparisons & alternatives
Pick between the options.
- →AI Agent vs Trading Bot
A trading bot executes fixed rules deterministically; an AI agent interprets context and decides which steps to take — powerful for research, risky for execution.
- →Web2 Apps vs Web3 Apps
Web2 apps authenticate with accounts and store state on private servers; Web3 apps authenticate with a wallet signature and keep state on a public chain.
- →Single Agent vs Multi-Agent System
One well-equipped agent beats a crowd for most jobs; multi-agent wins on genuinely separable, parallel work.