Cardano
KLEAR3 ResearchPublished Updated
ADACardano, a proof-of-stake blockchain whose consensus protocol, Ouroboros, is built from peer-reviewed research, has run since September 2017. The repository most data sites link to for its code has sat archived since 2020; the one that actually maintains the network is younger and still active today.
What does Cardano actually do?
Cardano runs a proof-of-stake network secured by Ouroboros, the protocol Input Output built from peer-reviewed cryptography research rather than from Bitcoin’s energy-spending design. ADA, its native asset, pays transaction fees, is staked to help produce blocks, and carries voting weight in the on-chain governance that decides how the protocol itself changes.
Cardano describes itself, on its own homepage, as “a public blockchain bringing assurance to critical operations” — a narrower claim than “a faster Ethereum” or “a cheaper Bitcoin”, and one aimed at institutions that need a ledger they can audit rather than a network they can speculate on.
Source cardano.org, read 2026-10-14
The protocol that secures it, Ouroboros, is described by the project’s own documentation as “the first provably secure proof-of-stake protocol” and “the first blockchain protocol based on peer-reviewed research”. That lineage shows up in how the network changes: not by a company shipping an upgrade, but by a written proposal that has to survive review before it reaches a vote.
Source docs.cardano.org, read 2026-10-14
This page does not examine Cardano’s smart-contract layer, Plutus, or any application built on top of it. It covers the chain itself: who maintains its code, how its token moves, and what its governance can and cannot decide.
Is the code actually open?
Published, yes, and it is easy to read the wrong repository. CoinGecko’s own link list names input-output-hk/cardano-sl first, archived since 2020, before the actively maintained client under the Apache-2.0 licence that currently takes 236 commits in 90 days from 18 distinct identities.
input-output-hk/cardano-sl holds 3 743 stars, and has not taken a commit since 30 July 2020. It is the code of the Byron era, written before the Shelley rewrite, and it is the first of five repositories CoinGecko lists for Cardano on its own project page. A reader who clicks that link sees real, readable, star-heavy code — and none of it has moved in six years.
Source GitHub API, read 2026-10-14
The repository that actually ships the software running the network today is IntersectMBO/cardano-node, opened on 23 May 2019 under the Apache-2.0 licence. It carries fewer stars than the frozen one it replaced — popularity measures a project’s whole history, not its current state.
Source GitHub, read 2026-10-14
No dedicated repository of third-party audit reports is published for the node, unlike the security-audits repositories Solana and Ethereum each keep. A bug bounty programme exists; an audit report is a different thing, and we did not find one.
Source GitHub Security — IntersectMBO/cardano-node, read 2026-10-14
Is anyone still working on it?
Yes. The maintained client, IntersectMBO/cardano-node, took 236 commits from 18 distinct Git identities in the 90 days to 14 October 2026, under a licence and a repository the archived one never carried. The proposal process behind the chain has run even longer: open since April 2020, 771 pull requests merged.
That commit count is a window, not a total, and it moves on its own: the fenced period ends the day we read it and begins exactly 90 days earlier. Eighteen Git identities are not eighteen people — one contributor can sign with several, and several can share one.
Source GitHub, read 2026-10-14
The repository that decides how Cardano itself changes is older than the node client and just as active. cardano-foundation/CIPs has merged 771 pull requests since 2020-04-09, publishing 158 numbered proposals so far — the written record behind every change this page describes.
Source cardano-foundation/CIPs, read 2026-10-14
What shows it is running
- The chain responds — block 14,066,776, epoch 661
- The maintained client is still being worked on — 236 commits in the 90 days to 2026-10-14, 18 distinct Git identities
Source api.koios.rest · GitHub, read 2026-10-14
- Cardano mainnet goes live (Byron)
- The maintained client opens
- The repository most sites still link to is archived
- On-chain governance activates (Chang)
- This page, checked and published
Where does it actually run?
On its own chain, and we checked it responds: a public indexer returned block 14 066 776 in epoch 661 on 14 October 2026. Cardano runs no virtual machine shared with another network’s tokens; what settles here is ADA itself, staked to the pools that keep the chain moving.
That is a narrower claim than most of the chains in this series make. There is no dollar token bridged in from elsewhere to track, no deposit figure from an aggregator to disclaim: a single network, a single asset, measured directly.
Source api.koios.rest, read 2026-10-14
Its documentation site answers normally, which matters here because the governance record, the tokenomics page, and the hard-fork history this page cites all live there rather than in a whitepaper frozen in 2017.
Source docs.cardano.org, read 2026-10-14
Network verified by reading its current block height on a public node.
What the token actually does
ADA pays fees, is staked by pools to produce blocks, and since the Chang upgrade carries a vote: delegate it to a registered representative, or register as one, and it weighs on which governance actions pass. Supply is capped at 45 billion; 37.55 billion already circulate.
Every transaction fee, plus a fixed 0.3% of whatever reserve remains each epoch, lands in a shared pot. 20% of that pot goes to the treasury before the rest pays the pools and delegators who produced the block.
Source docs.cardano.org, read 2026-10-14
That 0.3% figure is described, in the project’s own documentation, as a reserve 4-to-5-year half-life — not a cliff, a slow decline. Total supply is capped at 45 billion ADA, of which 37.55 billion already circulates, a figure read from CoinGecko rather than from the chain itself.
Source CoinGecko, read 2026-10-14
Governance weight came later than the token did. The Chang hard fork, on 2024-09-01, enabled only parameter changes and hard fork initiations; the full set of governance actions and the Delegate Representative role itself only went live with Plomin, on 2025-01-29.
Source cardano.org/hardforks, read 2026-10-14
- In circulation37,553,537,941.909
- Issued, not circulating7,446,462,058.091
- What the token is forDirectly verified
- How many existThird-party supported
- How new ones appearDirectly verified
- Who can change the rulesDirectly verified
The KLEAR3 rating
4.2out of 5 in the KLEAR3 rating
Each axis gets 0 to 5 from the facts in the register, against thresholds written in advance. The overall score is their weighted average, and it only appears once at least 5 of the 6 axes rest on a measured fact.
- Audits ×1.3
- Open code ×1.2
- Business model ×1.1
- Token alignment ×1
- Governance ×1
- Maturity ×0.7
- Open code5
public repository · Apache-2.0 licence · 103 published releases · 236 commits in 90 days
- Audits3
code reviewed in the open: 236 commits in 90 days, 7 years in production
What holds it back no report in our registry · no report opened: we do not know what they found
- Business model5
business model published by the protocol · paid by whoever sends the transaction, plus the remaining reserves at each epoch · collected by pool operators and their delegators · 20% of the pot goes to the treasury each epoch; the rest funds that epoch’s rewards
- Governance5
decision mechanism published · 158 improvement proposals published · process open for 6 years
- Token alignment3
supply capped at 45 billion · 83% already in circulation · emission schedule published
What holds it back fate of the initial allocations not established
- Maturity4
7 years of public code · 1 network verified by us
We rate the protocol: how it is built, how much of it can be checked, how its token lines up. Never an investment. A high score is not advice to buy, and a low one is not advice to sell.
What we could not establish
How the first ADA were divided among the organisations that built Cardano before mainnet launch, how concentrated Delegate Representative voting power actually is today, and whether any third-party audit of the node exists outside a bug bounty programme. We read the documentation and the repositories, not every arrangement around them.
Cardano is not built by an anonymous team: Intersect is a named, member-based organisation with its own public repositories, and it says so on its own site. What it does not publish, in anything we read, is how much voting weight sits with any single Delegate Representative or pool operator today — the question that decides whether “two of three bodies must agree” means a real check or a formality.
Source intersectmbo.org, read 2026-10-14
The genesis distribution — what the founding organisations held before anyone could buy ADA on an exchange — is a question this campaign did not answer. It pre-dates the repository we measured, and a registry that has not read a source does not guess at it.
Frequently asked questions
Why does the repository with the most stars show no recent activity?
Yes, and for a specific reason: data aggregators list input-output-hk/cardano-sl first, and it has taken no commit since July 2020. The network itself did not stop; the team moved the active codebase to a different repository, under a different organisation, five years ago.
IntersectMBO/cardano-node, the client that actually produces blocks today, opened in May 2019 and still carries fewer stars than the frozen one it replaced. Stars accumulate over a project’s whole history and never fall; the date of the last commit is the only field that answers what a reader actually wants to know, and the archived repository shows that date plainly if anyone checks it.
This is the same trap documented on Solana, where the archived solana-labs/solana still outranks the maintained anza-xyz/agave in visibility. A link that loads, shows real code, and never answers 404 is exactly what makes an archived repository convincing.
Does on-chain governance mean ADA holders already control the treasury?
Not yet in full. The Chang hard fork, 1 September 2024, enabled only parameter changes and hard fork initiations under CIP-1694. The full set of governance actions, and the Delegate Representative role itself, only went live with Plomin, 29 January 2025.
Since Plomin, a governance action still needs ratification from at least two of three bodies: Delegate Representatives chosen by ada holders, the Constitutional Committee, and the stake pool operators who already produce every block. Nobody we read assigns a single body full control, and we have not measured how concentrated the DRep vote actually is.
The improvement-proposal repository behind all of this has merged 771 pull requests since April 2020, which is the part of governance that was never gated behind a hard fork: writing a CIP and arguing it through review has been open the whole time.
Is Cardano’s ADA supply capped?
Yes: 45 billion, read from CoinGecko rather than our own chain query, with 37.55 billion already circulating. New ADA still enters through monetary expansion rather than mining: a fixed 0.3% of the remaining reserve each epoch, which the documentation itself describes as a four-to-five-year half-life.
Twenty percent of each epoch’s combined pot, fees plus that expansion, goes to the treasury before anything reaches a pool or its delegators. That split funds the ecosystem without a separate tax, and it is the same 20% whether the ada came from a transaction fee or from the reserve.
What we did not establish is how the first ada were divided among the organisations that built Cardano before mainnet launch. That history pre-dates the repository we measured, and a registry that has not read it does not guess at it.
Where to read on, at the source — every address answered 200 on 2026-10-14.
Next: our NEAR profile shows another proof-of-stake chain whose GitHub licence badge contradicts its own files, and our methodology sets out what counts as a source here.
Ask an assistant to summarise it and check what it claims.