Web3 Case Study Role Misconfig logo

Web3 Case Study Role Misconfig

CommunityPopular
tradecatlabs
web3-case-study-role-misconfig

Case study - role misconfiguration bug class applied to a yield aggregator protocol. Use as a template for applying all 10 bug classes to a single target.

Overview

Publishertradecatlabs
Repositoryvibe-coding-cn
Skill nameweb3-case-study-role-misconfig
Stars
16.3K
Forks
1.6K
Bundled files
Instructions only
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by tradecatlabs on GitHub. Read the source before you install it.

Installation

Install the Web3 Case Study Role Misconfig AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/tradecatlabs/vibe-coding-cn.git /tmp/vibe-coding-cn
mkdir -p .claude/skills
cp -r /tmp/vibe-coding-cn/research/vibe-cybersecurity-cn/skills/web3-bug-bounty-hunting/web3-case-study-role-misconfig .claude/skills/web3-case-study-role-misconfig
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Web3 Case Study Role Misconfig in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Web3 Case Study Role Misconfig on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Web3 Case Study Role Misconfig is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

CASE STUDY: ROLE MISCONFIGURATION IN A YIELD AGGREGATOR

Bug Class: Access Control | Severity: Critical/Medium | Payout Range: $10K–$50K This file shows how to apply the full 10-class methodology to a real yield aggregator target.


TARGET PROFILE (Anonymized)

FieldValue
Protocol TypeYield aggregator — stablecoin → lending protocol → harvest → DEX → reward token
Max Bounty$50K (Critical)
TVLLow (fresh program, under $100K)
Core ContractsVault.sol, RewardsDistributor.sol
Program Age~5 days when hunted (fresh = low competition)
Prior AuditsFirm A (16 findings, all Risk Accepted) + Firm B (18 findings, all Risk Accepted)

Scorecard: Max bounty (+2) + custom math (+1) + recent code (+1) + known prior audits (+1) + public source (+1) + program new (+2) = 8/10 → HUNT

Why this scores high: Fresh program on a live bounty platform + prior audits that accepted all risk = team is aware of issues but hasn't patched them. Hunt for what auditors missed or flagged but accepted.


ARCHITECTURE + FUND FLOW

User deposits Stablecoin
    ↓ deposit(uint256 amount)
Vault.sol stores:
  - deposits[user] += amount
  - totalDeposited += amount
  - depositTimestamp[user] = block.timestamp
    ↓ safeTransferFrom(user, address(this), amount)
    ↓ lendingProtocol.supply(stablecoin, amount, address(this), 0)
  Interest-bearing token accrues in Vault.sol balance
    ↓ (periodic) _performHarvest()
  aToken balance > totalDeposited + DUST_THRESHOLD
    ↓ lendingProtocol.withdraw(stablecoin, harvestAmount - 1, address(this))
    ↓ dex.exactInputSingle(stablecoin → rewardToken)
    ↓ RewardsDistributor.distribute(rewardToken, amount)
  RewardsDistributor tracks:
  - cumulativeRewardPerShare updates
  - users can call claimFor(user) to collect rewardToken

User withdraws:
    ↓ withdraw(uint256 amount)
  if block.timestamp < depositTimestamp[user] + LOCK_PERIOD:
    withdrawFee applies (e.g. 0.5%)
  lendingProtocol.withdraw(stablecoin, amount, user)

Key state variables:

  • deposits[user] — user principal (stablecoin)
  • totalDeposited — sum of all principals
  • depositTimestamp[user] — last deposit time (affects withdrawal fee)
  • cumulativeRewardPerShare — reward index in RewardsDistributor
  • lastClaimedReward[user] — user's last reward index

KNOWN ISSUES (Risk Accepted by Team — Do NOT Submit)

Firm A Findings (16 total, all Risk Accepted)

All standard: missing events, gas optimizations, reentrancy guards present (CEI followed), centralization risks (owner can pause), single oracle (DEX swap is operational, not security-critical).

Firm B Findings (18 total, all Risk Accepted)

Including:

  • HAL-01: withdrawFee can be changed by owner (centralization)
  • HAL-05: deposit() resets depositTimestamp even on partial top-ups → extends lock period for existing deposits
  • HAL-08: Missing check for DISTRIBUTOR_ROLE being set (flagged but did NOT verify it was never granted)
  • Various gas and event issues

Pattern: Firm B flagged "missing check" but didn't verify the role was actually ungranted. This is the gap to exploit.


BUG CLASS VERDICTS

1. Accounting Desync — 2 FINDINGS

Finding 1: The -1 Stranding Pattern

solidity
// In _performHarvest():
harvestAmount = aToken.balanceOf(address(this)) - totalDeposited - 1; // strands 1 wei

The hardcoded -1 strands 1 wei of stablecoin per harvest permanently. Over thousands of harvests, this accumulates. Severity: LOW/INFORMATIONAL (no user loss, just protocol dust accumulation).

Finding 2: Dust Harvest DoS ← VALID MEDIUM

Scenario: Accumulated harvest amount is very tiny (< DEX minimum swap)
1. harvest() calls dex.exactInputSingle(stablecoin → rewardToken)
2. DEX returns 0 (amount too small to produce any output)
3. RewardsDistributor.distribute(0) is called
4. If distribute() reverts on 0 amount → harvest is permanently frozen
5. Users can still withdraw principal but all future yield is lost

Verification: Check if distribute(0) reverts. Check DEX minimum swap threshold.

2. Access Control — 1 FINDING (CRITICAL/HIGH)

Finding: DISTRIBUTOR_ROLE Never Granted ← MAIN FINDING

solidity
// RewardsDistributor.sol
bytes32 public constant DISTRIBUTOR_ROLE = keccak256("DISTRIBUTOR_ROLE");

function claimFor(address user) external {
    require(hasRole(DISTRIBUTOR_ROLE, msg.sender), "Not distributor");
    // ... distribute rewardToken to user
}

Problem: DISTRIBUTOR_ROLE is defined but NEVER granted in the constructor or any initialization function. No address holds this role. claimFor() can never succeed — all reward tokens are permanently locked.

How Firm B missed it: They flagged "missing check for whether role is set" — but their fix recommendation was "add a require that checks the role exists." They didn't verify that getRoleMemberCount(DISTRIBUTOR_ROLE) == 0 on the live deployment.

Severity Assessment:

  • If harvest HAS already happened: CRITICAL (funds locked forever)
  • If harvest never happened yet: HIGH (permanent lock when it does happen)
  • Impact × Likelihood × Exploitability: 3 × 3 × 3 = 27 → CRITICAL

Verification commands:

bash
# Check if any address has DISTRIBUTOR_ROLE (replace with actual address)
cast call <REWARDS_DISTRIBUTOR_ADDR> \
  "getRoleMemberCount(bytes32)(uint256)" \
  "$(cast keccak 'DISTRIBUTOR_ROLE')"
# Expected: 0 = confirmed bug

# Alternative: Etherscan → Events → filter "RoleGranted"
# If no RoleGranted events with DISTRIBUTOR_ROLE hash = confirmed

3. Incomplete Path — Known (Risk Accepted)

Firm B HAL-05: deposit() resets depositTimestamp[user] even on partial top-ups, extending the lock period for all existing deposits. Risk Accepted by team.

4. Off-by-One — CLEAN

All boundary operators (>=, <) in Vault.sol and RewardsDistributor.sol are correct.

5. Oracle Price — CLEAN

Protocol does NOT use price oracles for security decisions (no lending, no liquidation, no collateral). The DEX swap is operational (converting yield), not security-critical. MEV/sandwich risk exists but is a griefing/efficiency issue, not a theft vulnerability.

6. ERC4626 Vaults — NOT APPLICABLE

Uses a custom 1:1 share model, NOT ERC4626:

  • deposits[user] tracks exact principal
  • No share price, no share-based rounding
  • Transfers between users are blocked
  • First depositor inflation attack does NOT apply

7. Reentrancy — CLEAN

Follows CEI (Checks-Effects-Interactions) correctly:

  • deposits[user] += amount BEFORE lendingProtocol.supply()
  • deposits[user] -= amount BEFORE lendingProtocol.withdraw()
  • Missing nonReentrant guard, but CEI makes it safe. Not submittable without PoC.

8. Flash Loan — CLEAN (Economically)

Flash loan attack would attempt: deposit → dilute harvest → withdraw to steal yield. The withdrawFee makes this unprofitable:

  • Attacker deposits $1M → harvest dilutes → attacker gains $0 extra yield
  • But: attacker pays withdrawal fee to exit
  • Net: negative expected value → NOT PROFITABLE

9. Signature Replay — NOT APPLICABLE

No signature-based functions, no EIP-2612 permit, no meta-transactions.

10. Proxy/Upgrade — NOT APPLICABLE

Not upgradeable proxies. No proxy pattern.


WHAT TO SUBMIT

SUBMIT (2 findings):

Finding 1 — CRITICAL/HIGH:

Title: DISTRIBUTOR_ROLE never granted in RewardsDistributor.sol,
       permanently locking all reward tokens

Root Cause: DISTRIBUTOR_ROLE is defined but grantRole() is never called
            in constructor or initialization. No address holds this role.

Impact: All rewards distributed by harvest() are permanently locked —
        claimFor() always reverts.
        Users receive zero yield despite depositing and paying withdrawFee.

Attack Path: Not an attack — passive failure. Any harvest → rewards locked forever.

Severity: Critical (if harvest has occurred) / High (if not yet)

Finding 2 — MEDIUM:

Title: _performHarvest() dust harvest causes permanent DoS on yield distribution

Root Cause: When accumulated yield rounds to < DEX minimum swap amount,
            exactInputSingle() returns 0 output. distribute(0) may revert or
            permanently advance the reward index with no rewards.

Impact: After a dust harvest, all subsequent harvests may fail permanently.

Severity: Medium (requires specific conditions but permanently impacts yield)

DO NOT SUBMIT:

  • The -1 stranding (informational, design choice)
  • depositTimestamp reset (Risk Accepted by team)
  • Missing nonReentrant (CEI is followed; no PoC = no submission)
  • Owner centralization (excluded by design in Immunefi SC programs)
  • Any of the already-acknowledged findings from prior audits

COMPLETE POC TEMPLATE: ROLE NEVER GRANTED

solidity
// SPDX-License-Identifier: UNLICENSED
pragma solidity ^0.8.13;
import "forge-std/Test.sol";
import "forge-std/console.sol";

interface IRewardsDistributor {
    function DISTRIBUTOR_ROLE() external view returns (bytes32);
    function hasRole(bytes32 role, address account) external view returns (bool);
    function getRoleMemberCount(bytes32 role) external view returns (uint256);
    function claimFor(address user) external;
}

contract RoleNeverGrantedTest is Test {
    // Replace with actual deployed address from target
    address constant REWARDS_DISTRIBUTOR = address(0xYOUR_TARGET_ADDRESS);

    IRewardsDistributor distributor;

    function setUp() public {
        // Fork at current block
        vm.createSelectFork(vm.envString("MAINNET_RPC_URL"));
        distributor = IRewardsDistributor(REWARDS_DISTRIBUTOR);
    }

    function testDistributorRoleNeverGranted() public {
        bytes32 DISTRIBUTOR_ROLE = distributor.DISTRIBUTOR_ROLE();

        uint256 memberCount = distributor.getRoleMemberCount(DISTRIBUTOR_ROLE);
        console.log("Addresses with DISTRIBUTOR_ROLE:", memberCount);

        // Should be 0 — proving no one can call claimFor()
        assertEq(memberCount, 0, "DISTRIBUTOR_ROLE has 0 members (confirmed bug)");

        // Verify claimFor() reverts for any user
        address testUser = address(0x1234);
        vm.expectRevert(); // "AccessControl: account does not have role"
        distributor.claimFor(testUser);

        console.log("CONFIRMED: claimFor() reverts for all users.");
        console.log("All rewards permanently locked.");
    }
}

Run:

bash
forge test --match-test testDistributorRoleNeverGranted -vvvv \
  --fork-url $MAINNET_RPC_URL

Expected output:

Addresses with DISTRIBUTOR_ROLE: 0
CONFIRMED: claimFor() reverts for all users.
All rewards permanently locked.

[PASS] testDistributorRoleNeverGranted()

REPORT TEMPLATE

Title: Missing DISTRIBUTOR_ROLE grant permanently locks all rewards for all users

Bug Description: RewardsDistributor.sol defines DISTRIBUTOR_ROLE and requires it to call claimFor(). However, grantRole(DISTRIBUTOR_ROLE, ...) is never called in the constructor, any initialization function, or any privileged setter. No address holds this role. claimFor() always reverts.

The Vault contract calls RewardsDistributor.distribute() after each harvest, successfully depositing reward tokens into the distributor. However, these tokens can never be claimed — permanently locked.

Root Cause: Constructor is missing grantRole(DISTRIBUTOR_ROLE, vaultContract).

Impact: All yield earned by all depositors is permanently locked in RewardsDistributor.sol. Users cannot receive any return from their deposits.

Impact category: Permanent freezing of funds

Proof of Concept: [Run PoC above against deployed contract]

Remediation: Add grantRole(DISTRIBUTOR_ROLE, address(vault)) in RewardsDistributor constructor, OR add a setDistributor(address) function callable only by admin.


VALIDATION CHECKLIST (7-Question Gate)

QuestionDISTRIBUTOR_ROLEDust Harvest DoS
Can attacker use it NOW?Yes (passive — already locked)Yes (needs small harvest)
Impact in program's list?Permanent freezing of funds ✓Temporary/permanent DoS ✓
In-scope contract?RewardsDistributor.sol ✓Vault.sol ✓
Requires admin access?No — passive ✓No — anyone can trigger ✓
Already known/acknowledged?No ✓No ✓
Economically viable?Yes — passive lock ✓Yes — dust accumulates ✓
Already public?No ✓No ✓
VERDICTSUBMITSUBMIT

LESSONS: WHAT TO LOOK FOR ON SIMILAR TARGETS

This pattern (role defined, never granted) appears frequently in:

  1. Any contract using OpenZeppelin AccessControl — check every bytes32 public constant *_ROLE — verify each one has a corresponding grantRole() call in constructor
  2. Distributor/reward contracts — these are often deployed separately and the grant is "supposed to happen" during deployment setup
  3. Contracts with multiple initialization steps — if setup requires calling multiple functions in sequence, the grant is often missed

Grep to find candidates:

bash
# Find all role definitions
grep -r "bytes32 public constant.*ROLE" ./src/

# Verify each role has a grantRole call
grep -r "grantRole" ./src/

# If roles > grantRole calls → investigate each missing grant

Protocols to check for this pattern: Any protocol where RewardsDistributor, FeeDistributor, YieldDistributor, or Distributor is a separate contract from the main vault.


→ NEXT: 08-ai-tools.md

Frequently asked questions

What does the Web3 Case Study Role Misconfig AI skill do?

Case study - role misconfiguration bug class applied to a yield aggregator protocol. Use as a template for applying all 10 bug classes to a single target.

Why use Web3 Case Study Role Misconfig on TypingMind?

Because you install it once and use it with any model. Web3 Case Study Role Misconfig is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Web3 Case Study Role Misconfig in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/tradecatlabs/vibe-coding-cn/tree/develop/research/vibe-cybersecurity-cn/skills/web3-bug-bounty-hunting/web3-case-study-role-misconfig. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Web3 Case Study Role Misconfig?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Web3 Case Study Role Misconfig?

As many as you like. As long as a model supports skills, you can use Web3 Case Study Role Misconfig with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Web3 Case Study Role Misconfig AI skill free?

Yes. It is published on GitHub by tradecatlabs under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇