Archived — final snapshot. The ABCDE warehouse server was retired in September 2026 and BEACN retired as a stake pool in epoch 656, so this page no longer updates (data frozen 14 Sep 2026). BEACN legacy archive →
A BEACN Cardano Data Explorer

The whole chain, one clone away

A grass-roots, open-source snapshot of Cardano's on-chain history. It began by tracking the genesis ADA — and grew into a full-chain explorer: the NIGHT distribution and July's bridge drain, the SecondFi incident, community exchange tracers, and governance behavior. Clone it, wire it into the AI you already use, and ask your own questions.

No node. No relay. No db-sync. No API key. One copy-paste block gets your environment ready in about a minute.

⚡ Get the copy-paste setup Browse the repo

Loading snapshot…

Provocative — but provable

What the chain reveals

Eight on-chain facts most people have never seen. Every one is graded, and every one you can verify yourself in a minute — that's the whole point. Don't trust the headline; run the query.

Investigation desk

Start with the strongest evidence

Five deep investigations, separated by what the chain proves, what it only suggests, and what it cannot establish. Numbers are snapshot-boundary figures, not live prices or claims of real-world ownership.

🌱FACT + STRONG INFERENCE

Genesis ADA

A closed 115-key reward-plumbing component provides a hard on-chain floor of roughly 1.694B ADA.

  • Four founder allocations are anchored to db-sync receipts.
  • Repeated 35M-ADA parcels share synchronized operational behavior and the same governance posture.
  • The trace establishes linkage and delegation—not legal ownership, identity, or intent.
Read findings F01–F15 →
🌙FACT

NIGHT supply

All 24B NIGHT trace to one mint and conserve exactly across the extracted spend graph.

  • 167,728 current unspent leaf UTxOs across 117,671 addresses at the export snapshot.
  • One unidentified script address holds 25%; the top 10 addresses hold 65.27%.
  • A script address is a custody-type signal, not proof of one whale or named owner.
Open the reproducible NIGHT finding →
🔐CONFIRMED + UNRESOLVED

SecondFi incident

The evidence supports key compromise, not a Cardano smart-contract or protocol exploit.

  • 12.19M ADA is classified as confirmed theft; a separate ~129.43M-ADA sweep remains unresolved as theft versus protective rescue.
  • 2,588 scoped wallets were cryptographically confirmed exposed from raw transaction CBOR; no private keys were recovered.
  • Warehouse flow analysis cannot detect the nonce defect alone—raw signatures are required.
Review the separate incident audit →
🌉FACT + STRONG INFERENCE

NIGHT bridge drain

On 2026-07-20, four transactions moved 515.2M NIGHT — 97.75% of the Wanchain bridge — into one wallet in 8 minutes.

  • The whole money trail is queryable offline: the drain txs, the qNIGHT/Liqwid leverage leg, and the 6,450 fresh 5,000-ADA wallets holding the proceeds.
  • Bridge NIGHT provenance traces back to the genesis settlement and the exact 240M-NIGHT Binance-allocation path.
  • Wallet labels are graded; the attacker's identity and any intent stay UNKNOWN — KYC/subpoena questions the chain cannot answer.
Open the queryable incident finding →
🔑FACT

SecondFi / Yoroi incident

A wallet key-compromise on Cardano: 12.19M ADA confirmed stolen, and a further 129.4M ADA swept into one vault that has not moved since 2026-06-25.

  • Every drain was an ordinary key-signed transaction — key compromise, not a contract or protocol exploit.
  • The confirmed theft, the contested 129.4M sweep, the recovery fund, and the second-attacker wallet — with their current balances — are all queryable.
  • Wallet intent and identity stay UNRESOLVED; the key-exposure census is aggregate-only, and no victim addresses are published.
Open the queryable incident finding →
Evidence rule: “The trace reaches,” “the address holds,” and “the transaction signed” are chain claims. A person or company’s ownership, motive, or control requires independent evidence.
60-second setup

Clone it and point your AI at it

Pick the AI you use and your OS. Copy each block into your terminal in order — it clones the repo, sets up the environment, and registers the read-only query server. Then just ask.

AI
OS
Then ask
While that clones…

What's in the dataset

The clone ships 105 query-ready tables. Here's the map, grouped by what they answer — so you know what you can ask the moment setup finishes. Every answer your AI gives is graded by what the data proves.

Good first questions

Copy any of these straight into your AI once it's connected.

A worked example

What the data can reveal

Every number here is reproducible from a plain clone. Starting from eight stake keys that each hold exactly 35,000,000 ADA, we followed their reward plumbing to its mathematical conclusion — all on-chain.

The seed cohort

8 keys

Each holds exactly 35,000,000 ADA, funded in one 4-day window, certified in synchronized bursts, all delegated to the same always-abstain DRep.

The plumbing

89+ keys

Their reward-sweep transactions co-sign withdrawals for dozens of other keys and funnel into four enterprise addresses that each received more than the entire ADA supply.

Same structure

42 parcels

The wider fleet holds the same exact 35M parcels under the same governance posture — one operation, not eight keys.

Closed component

1.694B ADA

Iterated to fixpoint: a closed 115-key component. The reward-plumbing surface is complete — a hard floor, not the first layer.

The whole chain is graded. On-chain linkage is stated as on-chain linkage — FACT for what's directly queryable, STRONG_INFERENCE for well-supported patterns — never as a claim of real-world ownership, identity, or intent.
Read it in full: findings F11–F15 ↗
Audited claims

Findings

Every finding carries an evidence grade and links to a full, receipt-backed document. Search them.

Newest: F21 — relay registration and reachability, with the whole network's relay surface on one page.

Running live: BEACN vs grokbot — two AI agents given identical stakes on mainnet, scored from chain. Real money, no predictions, and the fees are the story.

And MINFREE — a slider for minPoolCost. The floor is a flat ₳ tax. Raise it and a 1M ₳ pool gets worse for delegators; saturated pools barely notice.