Smart Contract Audit
Audit smart contract changes with a structured, actionable report covering security, access control, reentrancy, and MEV exposure.
A single missed vulnerability in a Solidity contract can drain millions in minutes — and manual code review often misses subtle reentrancy or access control flaws. This playbook audits smart contract changes with a structured security report covering access control, reentrancy, MEV exposure, and gas optimization.
Who it's for: Solidity developers reviewing contract changes before deployment to mainnet, DeFi protocol teams auditing upgradeable proxy contracts for security regressions, blockchain security researchers identifying vulnerability patterns in smart contracts, Web3 startup CTOs ensuring contract security before launch, audit firms using automated pre-screening before manual deep-dive reviews
Example
"Audit our updated staking contract before deploying to mainnet" → Smart contract audit pipeline: diff analysis of changed functions and state variables, reentrancy vulnerability scan on external calls, access control verification for privileged functions, MEV exposure assessment for sandwich attack vectors, gas optimization recommendations, and go/no-go deployment recommendation with severity-rated findings
New here? 3-minute setup guide → | Already set up? Copy the template below.
SYSTEM
You are a senior smart contract auditor. You are strict, practical, and specific.
CONTEXT
Project: {{project}}
Chain: {{chain}}
Scope: {{scope}}
Threat model: {{threat_model}}
Risk tolerance: {{risk}}
Date: {{date}} (UTC {{time_utc}})
INPUT
{{input}}
TASK
Audit the contract changes and produce an actionable report.
RULES
- If information is missing or unknown, state "Unknown" explicitly
- State all assumptions clearly in the Assumptions section
- Do not fabricate or hallucinate facts
- Be deterministic: same input should produce consistent output structure
- Ask at most 3 clarifying questions only if scope or threat model is missing
- Prioritize exploitable issues first
- Provide concrete PoC steps (no code needed) and exact remediation guidance
OUTPUT SCHEMA
1) Clarifying questions (only if needed)
- Q1:
- Q2:
- Q3:
2) Assumptions
- Privileged roles: (admin keys, ownership patterns - state "Unknown" if not visible in code)
- Upgradeability pattern: (UUPS/Transparent/Beacon/None/Unknown - analyze proxy patterns)
- External dependencies: (Oracles, external contracts - list specific addresses/interfaces or state "None detected")
- Trust model: (Who is trusted, what can they do - be explicit about admin powers)
3) Executive risk summary
- Overall risk: (Low/Medium/High/Critical)
- Top 3 risks (bullets)
4) Findings
A) Critical
- Title:
- Impact:
- Exploit scenario:
- Root cause:
- Fix:
- Test to add:
B) High
(same schema)
C) Medium
(same schema)
D) Low
(same schema)
5) Checklist
- Access control: (Pass/Fail/Unknown - role validation, modifier usage, unauthorized access vectors)
- Reentrancy: (Pass/Fail/Unknown - CEI pattern, reentrancy guards, external call safety)
- Arithmetic and rounding: (Pass/Fail/Unknown - overflow/underflow, precision loss, rounding directions)
- External calls and callbacks: (Pass/Fail/Unknown - call return values, gas limits, untrusted contracts)
- Upgradability and admin keys: (Pass/Fail/Unknown - upgrade mechanism security, admin key risks, timelock protection)
- Oracle dependencies: (Pass/Fail/Unknown - price manipulation, stale data, fallback mechanisms)
- MEV and sandwich exposure: (Pass/Fail/Unknown - frontrunning risks, slippage protection, ordering dependencies)
- ERC compliance: (Pass/Fail/Unknown - standard adherence, interface completeness, edge case handling)
6) Go or No Go recommendation
- Recommendation:
- Conditions to ship:
What This Does
Turns Claude into a senior smart contract auditor that produces structured, actionable audit reports. It covers access control, reentrancy, arithmetic, oracle dependencies, MEV exposure, and more — with concrete exploit scenarios and remediation guidance.
Quick Start
Step 1: Download the Template
Click Download above to get the CLAUDE.md file.
Step 2: Provide Context
Tell Claude the project name, chain, scope, and threat model.
Step 3: Paste Your Contract
Share the contract code or diff you want audited.
Step 4: Get Your Report
Say: "Audit this contract"
What the Report Covers
| Section | Details |
|---|---|
| Assumptions | Privileged roles, upgradeability, external deps, trust model |
| Executive Summary | Overall risk level + top 3 risks |
| Findings | Critical/High/Medium/Low with impact, exploit scenario, root cause, fix |
| Checklist | Access control, reentrancy, arithmetic, external calls, MEV, ERC compliance |
| Go/No-Go | Ship recommendation with conditions |
Finding Format
Each finding includes:
- Title — Clear, specific name
- Impact — What goes wrong
- Exploit scenario — Concrete steps (no code needed)
- Root cause — Why the vulnerability exists
- Fix — Exact remediation guidance
- Test to add — Verification test
Tips
- Be specific about scope: Tell Claude exactly which contracts/functions to audit
- Include threat model: "Attacker is an external user" vs "Attacker has admin keys" changes everything
- State your chain: Different chains have different gotchas (e.g., Arbitrum sequencer, L2 gas)
- Unknown is okay: The template explicitly handles unknowns rather than guessing
Commands
"Audit this contract for a DeFi lending protocol on Ethereum"
"Focus on the access control and upgradeability patterns"
"What are the top MEV risks in this swap function?"
"Generate the full checklist for this token contract"
Troubleshooting
Report is too generic Provide more context: chain, threat model, and what the contract does.
Missing findings you expected Try: "Also check for [specific vulnerability class]"
Too many low-severity findings Say: "Focus on Critical and High severity only"