Portfolio / Projects / CrowdFund DAO
SMART CONTRACTS · WEB3 SECURITY

ACADEMIC TEAM PROJECT · 2026–2027 · BLOCKCHAIN & SECURITY

CrowdFund DAO: secure crowdfunding and token-weighted governance.

A decentralized crowdfunding dApp built by a three-person ENSA Beni Mellal team. Users create campaigns, contribute ETH, receive CFD voting power and vote on fund releases. The contracts manage refunds, payouts and the security controls around them.

SolidityOpenZeppelinHardhatERC-20 / ERC20VotesDAOReact / Ethers.jsIPFSSecurity testing

01 / CHALLENGE

Keep campaign funds under contract rules, not one wallet.

The dApp keeps the crowdfunding flow inside the contracts: campaign creation, ETH contributions, CFD minting, refunds and controlled withdrawals. A beneficiary cannot move campaign funds until the contract conditions and the governance decision are satisfied.

01Create a campaign

Goal, deadline, beneficiary and IPFS metadata CID.

02Contribute ETH

The crowdfunding contract accounts for funds and contributions.

03Receive CFD

The token contract mints 1,000 CFD per ETH contributed.

04Govern & settle

Vote on eligible proposals; perform controlled withdrawal or refund.

02 / ARCHITECTURE

What runs where.

The React frontend prepares contract calls with Ethers.js and MetaMask signs them. Crowdfunding.sol manages campaigns, ETH accounting and DAO decisions. CFDToken.sol handles token supply and voting checkpoints. IPFS stores campaign content while the contract keeps the metadata CID.

MetaMaskAccounts & signing React frontendEthers.js + ABI Crowdfunding.solETH escrow · campaigns · DAOMultisig approvals · pause control CFDToken.solERC20Votes · mint / burn IPFSImage · metadata · CID signcallsmint / burn / votesmetadata DAO vote: quorum + majority · High-value withdrawal: 2-of-3 signers
Important distinction: ETH is held by the crowdfunding contract; CFD is a separate governance token. CFD's 1,000:1 issuance rule is not an ETH redemption price. The IPFS CID identifies metadata content, not the money. The optional image is uploaded first, then a CIDv1 JSON document containing title, description and the image URI; only that metadata CID is recorded on-chain. The local Kubo node pins the uploaded objects for the demonstration, which is not evidence of public IPFS persistence.

03 / CONTRIBUTION & TEAM CREDIT

What I worked on.

My engineering work — Kossi Richard Allado

  • Implemented and tested CFDToken.sol: ERC-20 token, role-gated mint/burn, ERC20Permit/ERC20Votes integration and vote delegation behavior.
  • Implemented and tested Crowdfunding.sol: campaign creation, contributions, finalization, refund and withdrawal accounting.
  • Strengthened the accounting invariant: totalRaised = campaignBalances + campaignRefunded + campaignWithdrawn.
  • Built and exercised targeted tests for refund behavior and a recursive-call reentrancy attempt using an attacker contract.

Shared team deliverables

  • DAO proposals and token-weighted voting, including snapshots, quorum and majority.
  • Multisignature controls, emergency pause, wallet-connected React application and IPFS flow.
  • Hardhat integration and security verification; optional Python-based anomaly-signal checks.
  • Presentation, integration and group demonstration.

Team: ADJEVI Kodjo Marius-Ben, ALLADO Kossi Richard, KPADENOU Kossi Roland.

This was a three-person project. My public fork preserves the shared history and links back to the original team repository; this page separates my work from the parts delivered together.

04 / GOVERNANCE

The voting rules implemented in the contract.

CFD voting is part of the application, not Ethereum consensus. A proposal waits one day before voting opens and stays open for three days. Voting power is read from a historical snapshot; execution needs 20% participation and a strict FOR majority.

Example: 5,000 eligible CFD

At least 1,000 votes must participate to reach the 20% quorum. If 2,000 vote FOR and none AGAINST, quorum and majority conditions are met; execution still waits until the voting period closes.

DAO versus multisignature

The DAO records a collective decision weighted by eligible CFD votes. A payout above 10 ETH also requires approval from at least 2 of 3 designated trustees for the current withdrawal amount. The two mechanisms serve different purposes.

Release is a separate gate

Reaching a goal and finalizing a campaign as successful do not directly unlock funds. The beneficiary needs an executed Release proposal; before the deadline, the same rule permits an early release only after goal attainment and governance approval.

05 / DEMONSTRATION EVIDENCE

The local flow: contribution, vote and release.

These captures come from the local Hardhat demonstration. They show the campaign funded, the MetaMask contribution, the CFD vote and the approved Release proposal.

What these captures show: the wallet request, the funded campaign, the recorded vote and the approved proposal are separate steps. The funded-campaign capture still shows 0 ETH released; payout is a later beneficiary action after proposal execution.

06 / VALIDATION

Checks run on the project.

Control / testWhat was checked or reported
Reentrancy defenseTargeted attacker-contract attempt, state-before-transfer accounting, and guarded refund execution.
Fund accountingHistorical amount raised separated from remaining balance, withdrawals and refunds, with accounting invariants.
VotingSnapshot-based voting, one-day delay, three-day voting window, quorum and majority rules.
Emergency and payout controlsRole-gated pause can block contributions and ETH withdrawals; guarded ETH paths use reentrancy protection. Payouts above 10 ETH add the independent 2-of-3 trustee requirement.
Off-chain triage bonusRead-only Python analysis flags repeated amounts, potential linked addresses through shared campaigns and contribution spikes. These are transparent review signals, not automatic fraud findings or freeze decisions.
Hardhat20 contract and integration tests pass with npm test.
SlitherThe repository includes the static-analysis report and the documented review of its findings. Static analysis supports review; it does not replace a manual audit.
Python8 anomaly-analysis tests pass with npm run test:analysis.
Browser walkthroughThe browser flow covers image/metadata upload and reading, campaign creation, funding, voting, proposal execution, withdrawal, token unlock, finalization and refund.
Local servicesLocal demo stack: Hardhat chain ID 31337, Vite frontend on port 5173 and Kubo IPFS API on port 5001.
How to read these results: the tests verify the implemented scenarios and the static-analysis report helps identify review points. Neither one replaces an independent smart-contract audit.

07 / LIMITATIONS

Scope and next steps.

Deployment status

The current demonstration runs on a local Hardhat network. Advancing local blockchain time makes the one-day delay and three-day vote period practical to demonstrate during a classroom session. A public Sepolia deployment is the next deployment step.

Review and maintenance

The project still needs an independent external audit before any real-money use. Manual token unlocking after payment and the size of the main contract are the main areas to improve next.