agentleFS
iktok90-designiktok90-design/ai-smart-contract-auditor/skills/erc4626-inflation/SKILL.md

erc4626-inflation

Skill36 starsChanged 5 months agoMITView on GitHub

Detect ERC-4626 inflation/donation attacks — first depositor share-price manipulation, naive convertToShares math, missing virtual-shares defense. Activate on any ERC-4626 vault implementation, share/asset math, `convertToShares`, `convertToAssets`, `previewDeposit`, `previewMint`, `totalAssets`, `_decimalsOffset`.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If first depositor deposits a large amount (say, 10K USDC), donation impact is bounded.

    line 45
  2. Very specific:

    Use OZ ERC4626Upgradeable ≥ 4.9.0 with _decimalsOffset() = 6 (or higher for low-decimal assets).

    line 79
  3. Distinctive:

    Always validate round direction matches OZ semantics.

    line 83

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

ERC-4626 inflation attack detection

Background

The classic "donation" or "inflation" attack on share-based vaults. First depositor mints 1 wei share for 1 wei of asset → share price = 1. Attacker then donates large amount directly to the vault (bypassing deposit) → totalAssets balloons → share price massively inflated → next depositor rounds shares to 0.

Has caused real, public losses across many forks of naive ERC-4626.

When this applies

  • Any contract inheriting ERC-4626
  • Custom vaults with share/asset math
  • Yield aggregators
  • Lending markets that mint share-tokens

Detection patterns

Naive convertToShares math (CRITICAL)

function convertToShares(uint256 assets) public view returns (uint256) {
    if (totalSupply() == 0) return assets;
    return assets * totalSupply() / totalAssets();
}

First deposit: 1 share for 1 asset. Donation: totalAssets += 1e30. Next depositor's 1 asset → 1 * 1 / 1e30 = 0 shares. Funds lost.

No virtual-shares defense (HIGH)

Defense: OZ ERC4626 v4.9+ uses _decimalsOffset() to mint "virtual" shares:

function _convertToShares(uint256 assets, Math.Rounding rounding) internal view virtual override returns (uint256) {
    return assets.mulDiv(totalSupply() + 10 ** _decimalsOffset(), totalAssets() + 1, rounding);
}

This adds N virtual shares (typically 10**6 — 6 decimals offset) that nobody owns. Donations have far less leverage.

Dead-shares pattern not used (HIGH)

Alternative defense: mint a fixed amount (e.g. 1000 shares) to address(0) or address(this) on first deposit. Locks initial donation impact.

Minimum first deposit not enforced (MEDIUM-HIGH)

If first depositor deposits a large amount (say, 10K USDC), donation impact is bounded.

Read-only deposit pause until seed (HIGH)

Some impls require admin to seed vault before allowing deposits. Acceptable defense; centralized but bounded.

Asymmetric round-direction (HIGH)

ERC-4626 spec: rounding favors the vault.

  • deposit: shares round DOWN (favor vault)
  • mint: assets round UP (favor vault)
  • withdraw: shares round UP (favor vault)
  • redeem: assets round DOWN (favor vault)

Reversed direction = attacker can extract value via rounding.

Donation accounting after rebase (HIGH)

Vaults holding rebasing tokens (stETH, etc.) — totalAssets() increases via rebase, not via internal accounting. Attack vectors similar but defenseless naive impl.

totalAssets() returns IERC20.balanceOf(address(this)) (HIGH)

Naive implementation that always sums the actual balance is the vulnerable shape. Better: track internally and rebase carefully.

Severity rubric

PatternSeverity
Naive share math + no virtual shares + no dead-shares + no min-depositCritical
Naive math + minimum first deposit but very lowHigh
Asymmetric round direction (favoring depositor)High
totalAssets() = balanceOf(this) with rebasing token, no defenseHigh
Virtual shares of 0 decimals offset on 18-decimal asset (insufficient)High
Dead-shares but very small (e.g. 1 wei)High
OZ ERC4626 v4.9+ with default offset (6)Info (canonical defense)

Remediation patterns

  • Use OZ ERC4626Upgradeable ≥ 4.9.0 with _decimalsOffset() = 6 (or higher for low-decimal assets).
  • OR: mint 1000+ dead shares to address(0) on first deposit.
  • OR: require first deposit ≥ MIN (e.g. 1 unit of underlying).
  • OR: admin-seeded vault with no public deposits until seed.
  • Always validate round direction matches OZ semantics.

False-positive notes

  • A vault using OZ v4.9+ ERC4626 with explicit _decimalsOffset return is likely safe — verify the offset value is reasonable (6 for 18-decimal underlying is the canonical).
  • Custom vaults with admin-seeded model are defenseless against admin compromise but not against external donations.

Related

  • [[flash-loan-attacks]] §inflation
  • [[integer-issues]] §rounding
  • [[yield-aggregator-specialist]] (subagent)

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.