CryptoJake

KEYSTONE Investigation — Maker (pDAI) Leg: Case File Chain: PulseChain (chain 369, 0x171) Fork block: 17,233,000 (state+history copy of Ethereum; blocks ≤ this are copied mainnet, blocks > this are chain-369-native) Scope of this file: the MakerDAO / pDAI leg only. Compound (cTRL/CCTP), bridge, and Atropa legs are separate. Status: core mechanism node-verified end-to-end. Several downstream questions remain open (Section 6). Method note: every finding below is grounded in a named script run against a synced chain-369 node. Fork-gating (pre/post fork-block) and reading Maker’s own ledger (Vat/DaiJoin LogNotes) rather than inferring from ERC-20 transfers are the two controls that make the claims defensible. Executive summary Against fork-minted liquid-staking collateral (wstETH, rETH) copied into PulseChain’s Maker deployment, an operator drew ~3.4 billion pDAI post-fork through a programmatic Sybil fleet of ~2,000 urns, then fired an inherited Emergency Shutdown Module (ESM) to cage the Vat and strip legitimate governance — freezing the system with its own emergency brake. Tracing the drawn pDAI to where it was minted as an ERC-20 (via DaiJoin.exit) revealed the operation is larger than the wstETH leg alone: ~106 billion pDAI (gross) was exited to a single address, c108, alongside a dozen further billion-scale recipients. Those recipients are not liquidity pools — they are bespoke Maker-integrated contracts, operator EOAs, and at least one previously-unknown Vat-authorized contract. c108 (0xc108f4…df3) sits at the center: it fired the ESM, it is the largest single pDAI recipient, and it is tied to the distribution hub (KEYSTONE) and the fleet deployer (BOT) through shared infrastructure. Framing caveat (load-bearing): “drawn” and “exited” totals are gross throughput, not net holdings. pDAI cycles (exit → deploy → join → re-exit). Whether the 106B represents value held/extracted vs cycled as liquidity is not yet resolved (Section 6). This file deliberately avoids “theft”/“stolen” language pending that resolution. The mechanism, end-to-end (all steps node-verified) 1.Copied collateral. PulseChain copied Ethereum’s Maker state at the fork, including the wstETH-A / wstETH-B / rETH-A ilks and the liquid-staking token contracts. The forked staking tokens are mintable on the fork. 2.Post-fork draws. A fleet of ~2,000 urns each performed frob() on the Vat, drawing pDAI against the forked staking collateral. Fork-gated frob census confirms these draws occurred at blocks > 17,233,000 (i.e. genuine chain-369 activity, not copied debt). Net post-fork drawn: •WSTETH-A: ~2.74B pDAI •WSTETH-B: ~304M pDAI •RETH-A: ~361M pDAI •Total ≈ 3.4B pDAI against fork-minted staking collateral. 3.Sybil fan-out. The urns are not organic borrowers. In the frob data the per-urn dart values descend in a near-perfect ~100-unit sequence across dozens of distinct urns (e.g. WSTETH-B: 14,745,584 / 14,745,484 / 14,745,383 / 14,745,182 …). That signature is a single script minting urns in a loop, the small decrements being rate drift between blocks. 4.Self-credit. In every frob, the destination w, collateral source v, and urn u are the same address (w = v = u). The drawn vat.dai is credited to the urn itself — there is no routing at the frob layer. 5.ERC-20 conversion. Each urn calls DaiJoin.exit(usr, wad), burning internal vat.dai and minting pDAI ERC-20 to usr. This is where the money actually lands and where convergence appears (Section 4). 6.Governance capture + freeze. At block 23,225,241, c108 called fire() on the inherited ESM (0x09e05ff6…4c58). Because this ESM holds direct Vat auth, its fire() executed, in one transaction: deny(MCD_PAUSE_PROXY) (stripping MakerDAO’s legitimate governance executor) followed by cage() (freezing the Vat). This is why Vat.live = 0 while End.live = 1 — the cage bypassed the End module entirely.