Research lab · fixtures

What objectively happened

Reproduce and compare transaction-system behavior. This workbench does not pick routes or grant approvals.

Reproducible demos

No.	Case	Result
demo-01	Canonical Replay Synthetic ETH transfer. Fixture backend is SUPPORTED. Not a live archive replay.	MATCH
demo-02	Client Differential Same input, REVM-shaped fixture vs RPC-shaped fixture. Gas mismatch is classified D4.	MISMATCH
demo-03	Reorg Included on fork A-B-C-D, orphaned after A-B-C'-D'. BlockCore records objective lifecycle only. Assurance interpretation is Rethium.	MISMATCH
demo-04	Mainnet ETH transfer Mainnet tx 0x4ef15de6… in block 25969069. Prestate is parent-block snapshot from publicnode. REVM 40.0.3 is EXPERIMENTAL; fixture backend records the canonical receipt.	MATCH
demo-05	Contract COMPLETE gate (synthetic) Synthetic STOP bytecode with one recorded storage slot. Proves COMPLETE-only contract path. Gas calibrated to REVM 40.0.3 (21160). Not a mainnet ERC20.	MATCH
demo-06	ERC20 COMPLETE gate (synthetic) Synthetic token RETURN-true on transfer(address,uint256) calldata with COMPLETE storage. Not a mainnet ERC20. Not a real balance mutation.	MATCH

Research lab	Six fixtures. Demo 04, 05, and 06 are CI MATCH. Demo 06 is ERC20-shaped synthetic, not a mainnet token.
Wedge	Engine, fork, and client compatibility is the direction, not current capacity. One client (REVM). Corpus of six.
Not yet	No arbitrary tx-hash submit, mainnet ERC20 MATCH, Reth/Geth dual compare, or customer corpus upload.
Pin	Reth v2.5.2 · REVM 40.0.3 · 6 fixtures · 4 MATCH · 2 MISMATCH

BlockCore	Execution Truth · What is true? This workbench
Areth	Execution Intelligence · What is best / most robust? Separate repository
Rethium	Financial Execution Assurance · Was it allowed? Separate repository

Connectivity is not replay correctness. Correctness lives in Replay, Compare, Evidence, and CI artifacts.

Checking remote status