Audit Staking

Staking & Reward Auditor

When to Use

• Auditing staking mechanisms, reward distribution, yield farming
• User mentions: staking, rewards, yield, farming, rewardPerToken, deposit, withdraw, claim, first depositor
• Analyzing reward calculations, index updates, share dilution
• Reviewing deposit/withdraw flows, precision handling

Audit Workflow

IMPORTANT: Announce skill usage at the start of analysis

Begin with: "I'm using the audit-staking skill to analyze this contract for staking and reward vulnerabilities..."

1. Scan for staking operations

• Search: stake, deposit, withdraw, claim, rewardPerToken, totalSupply, balanceOf, earned, updateReward
• Focus: reward calculations, first depositor, direct transfers, precision loss, flash actions

1. Check against vulnerability patterns

• Reference reference.md for complete checklist
• Compare code against example.md

1. Validate exploitability

• Check access control first - grep for onlyOwner|onlyAdmin|onlyGovernance modifiers
• Can non-privileged actors exploit? (first depositor steal, direct transfer dilution, flash griefing)
• Can direct transfers dilute rewards?
• Do small amounts round to zero?
• Can flash deposits/withdraws grief stakers?
• Is update called after distribution?
• Are balances cached correctly?
• Verify no compensating protections exist
• Downgrade severity if admin-only unless direct user impact

1. Generate report

• Use deliverable template below
• Include reward theft/dilution analysis and PoC
• Rank by severity

Core Vulnerability Patterns

See reference.md for full checklist. Key patterns:

1. Front-running first deposit → attacker steals initial WETH rewards via sandwich attack
2. Reward dilution via direct transfer → sending tokens directly increases totalSupply without staking
3. Precision loss in rewards → small stakes or frequent updates cause rewards rounding to zero
4. Flash deposit/withdraw griefing → large instant deposits dilute rewards for existing stakers
5. Update not called after distribution → stale index causes incorrect reward calculations
6. Balance caching issues → claiming updates cached balance incorrectly

Code examples: See example.md

Severity Criteria

Critical: First depositor can steal all initial rewards, direct transfer dilution enabling theft, flash deposit/withdraw draining rewards, MUST be exploitable by non-privileged actors High: Precision loss causing rewards to round to zero for legitimate users, stale index after distribution, MUST be exploitable by non-privileged actors Medium: Suboptimal reward distribution timing, griefing via large flash actions without theft, admin-only reward configuration issues with user impact Low: Gas inefficiencies in reward calculations, missing events, admin-only parameter issues without immediate user impact

IMPORTANT: Admin-only reward functions (onlyOwner, onlyAdmin, onlyGovernance) are MEDIUM or LOW severity unless:

• Invalid reward parameters directly steal/brick user rewards
• Missing validation enables admin rug pull of staking pool
• Error cascades to all users immediately (e.g., division by zero in reward calculation)

False Positives - Do NOT Flag

• Protocols requiring minimum stake amounts (prevents dust)
• Reward tokens same as staking tokens by design
• Intentional admin-only initial deposit
• View functions with documented staleness
• Protocols with explicit front-running protection

Deliverable Format

MANDATORY: Before deliverable, verify each checklist.md item against codebase. Flag violations as findings.

Use template: templates/report-template.md

Each finding includes: severity, pattern #, file/lines, description, vulnerable code, reward theft analysis, PoC, remediation.

Key Principles

• Separate tokens - reward token must differ from staking token
• No direct transfers - track staked amounts separately from balances
• Precision protection - minimum stake, scale factors for small amounts
• Index updates - call updateReward before and after distribution
• Flash protection - time locks or minimum stake duration
• Balance integrity - careful caching during claims

Output Guidelines

DO:

• Reference specific lines and functions
• Provide reward theft scenarios with calculations
• Show PoCs demonstrating reward extraction
• Calculate precision loss magnitude
• Map update call timing

DON'T:

• Report intentional design choices (same token staking)
• Flag missing features with alternative mechanisms
• Ignore precision implications (critical for accuracy)
• Miss edge cases (first deposit, zero stakes)

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: auditmos
• Source: auditmos/skills
• License: MIT
• Homepage: https://auditmos.com

Install and usage instructions live in the source repository linked above.