What Matters
An “AI agent token” should be audited as a software project, financial instrument, governance system and onchain asset—not as a logo attached to an AI narrative. In 15 minutes, an investor can verify the contract, token rights, supply, unlocks, holders, treasury, code, product evidence, revenue claims and liquidity well enough to reject many weak projects.
DN Evidence Block
Reviewed: 20 November 2026. Evidence base: official project documentation, block explorers, GitHub repositories, issuer disclosures and SEC investor-risk guidance.
Boundary: a rapid audit is a rejection screen, not full due diligence or a price forecast. Contract, holder and market data can change after publication.
The Ten Evidence Checks
| Check | Question | Evidence | Reject when |
|---|---|---|---|
| 1. Contract identity | Is this the verified token? | Official docs, chain, explorer, proxy and decimals | Contract differs across sources or is unverified |
| 2. Token rights | What does ownership provide? | Utility, governance, fee, staking and legal terms | Only vague “ecosystem access” language |
| 3. Supply | What can exist? | Circulating, total, maximum and mint authority | Supply can expand without clear limits |
| 4. Unlocks | Who can sell and when? | Vesting contracts, schedules and recipient classes | Large near-term unlocks are hidden or discretionary |
| 5. Holders | How concentrated is control? | Top wallets, exchanges, treasury and bridges | Unexplained wallets control material supply |
| 6. Treasury | Who controls project assets? | Wallets, multisig signers, budgets and transactions | Single signer or undisclosed treasury |
| 7. Product | Does a working agent or service exist? | Live product, docs, users, transactions and uptime | Demo, waitlist or generated screenshots only |
| 8. Code | Is development real and deployable? | Repository, contributors, releases, tests and deployed versions | Forked code, no releases or unrelated commits |
| 9. Revenue | Does usage create economic value? | Onchain fees, invoices, audited accounts or reproducible dashboard | Trading volume called revenue or token emissions called income |
| 10. Liquidity | Can a position enter and exit? | Venue, depth, spread, pool ownership and market makers | Thin liquidity or removable issuer-owned pool |
Minute 1–2: Resolve the Contract
Start from official project documentation and compare at least one independent source. Record chain, contract address, symbol, decimals, token standard and whether a proxy or upgrade mechanism exists. A copied website can display a legitimate name while linking to a different contract.
On the explorer, inspect contract verification, deployer, creation transaction, ownership and mint or pause functions. If the token is upgradeable, identify the admin and timelock.
Minute 3: Define the Token’s Rights
Does the token pay fees, secure work, govern parameters, buy compute, provide access or only trade? Governance rights may be weak if insiders control quorum. “Staking” may be inflationary distribution rather than revenue.
Write one sentence: “Holding this token gives the owner…” If the documentation cannot complete the sentence precisely, the token utility is not auditable.
Minute 4–5: Supply and Unlocks
Compare circulating, total and maximum supply. Identify mint authority, emissions, burns and treasury allocations. Then map team, investor, ecosystem and community unlocks by date.
Do not calculate market capitalization from circulating supply alone. Fully diluted valuation and sellable supply after the next unlock can materially change the risk.
Minute 6: Holder Concentration
Label known exchange, bridge, liquidity, treasury, vesting and burn addresses before interpreting the top-holder list. A bridge contract can appear concentrated without one beneficial owner, while several related wallets can conceal common control.
Measure the percentage controlled by insiders and the liquidity available to absorb sales. Governance, treasury and market risk rise when one group dominates all three.
Minute 7: Treasury and Governance
Locate treasury wallets and inspect signers, timelocks and spending history. A public multisig is stronger evidence than a dashboard number. Review whether governance can mint tokens, change fees, upgrade contracts or move user assets.
Check whether proposals are debated and executed or whether voting exists only as marketing. Record emergency powers and who can use them.
Minute 8–10: Product and Users
Use the product. Confirm the agent, API, marketplace or protocol exists and performs the advertised job. Look for documentation, status history, transaction volume, paying customers and evidence of repeated use.
Followers, Discord members and wallet connections are not equivalent to customers. A token can have active trading while the product has no economic usage.
Minute 11–12: Code and Deployment
Inspect the official repository’s contributors, releases, tests, issues and dependencies. Compare the repository with deployed contracts or published package versions. A large commit count can be generated automatically or inherited from a fork.
For AI projects, identify the model, agent framework, data sources, tool permissions and what is actually decentralized. A token does not make a hosted API an autonomous network.
Minute 13: Revenue and Value Capture
Separate product revenue, protocol fees, trading volume, token emissions and treasury appreciation. Ask who pays, for what service, where the payment settles and whether token holders receive any economic benefit.
A high annualized staking yield funded by new token issuance is a distribution mechanism, not external revenue. Value capture should be documented rather than inferred from price.
Minute 14–15: Liquidity and Exit
Check reputable venues, pool depth, spread, 1% and 5% price impact, market-maker concentration and whether liquidity tokens can be removed. Review recent volume quality and large transfer behavior.
Do not use displayed price as valuation when the available exit size is small. Model the position at an executable price after fees and slippage.
DN Rule
Never buy an AI token until you can distinguish the product, the company or protocol, and the token. They can succeed or fail independently.
DN AI Agent Token Evidence Score
Score the evidence available now. Higher scores mean more unresolved risk.
Worked Example
A project claims to run autonomous trading agents and reports $20 million of “economic activity.” The contract is verified, but the token can be minted by one admin. The repository is a fork with few original releases. The $20 million figure is exchange trading volume, not customer revenue, and two unlabeled wallets hold 38% of supply.
The correct result is not “scam proven.” It is a high evidence-gap score: unclear value capture, concentrated authority, weak product-code linkage and major holder risk. That is enough to decline the investment until evidence improves.
Methodology and Limitations
DN designed a rapid rejection screen across contract, rights, supply, unlocks, holders, treasury, product, code, revenue and liquidity. It does not replace legal, security, financial or technical due diligence and cannot predict token price.
Frequently Asked Questions
Can a token project pass this audit and still lose value?
Yes. The checklist tests evidence and structural risk, not future price. Markets, regulation, competition and execution can still deteriorate.
Is GitHub activity proof of a real product?
No. Commits can be cosmetic, forked or unrelated to usage. Review contributors, releases, deployed code, tests and whether the product depends on the repository.
What matters most in tokenomics?
The rights the token grants, supply and unlock schedule, holder concentration, treasury control and whether product demand actually requires the token.
How do I verify the correct contract?
Use the project’s official documentation and multiple trusted sources, then compare chain, contract, decimals and proxy or upgrade status on a block explorer.
What is the fastest red flag?
A project that cannot clearly explain what token holders receive, who controls upgrades and treasury assets, or how reported revenue connects to the token.
Primary Sources
- Etherscan Information Center
- GitHub Documentation
- SEC Investor Alerts and Bulletins
- CoinGecko API Documentation
- OpenZeppelin Contracts Documentation
Editorial disclosure: no token project, exchange or data provider paid for inclusion or scoring.