01 / RECORD INTEGRITY
Committed history
Loading…Reading the current run.
SOURCE · RECORDS · RECOMPUTATION
Verify what you see.
Inspect committed inputs and neural activity, download the exact recording, and recompute it using its archived source. Computation, onchain anchoring and financial execution have different evidence.
02 / RECOMPUTATION
Neural replay
Loading…Replay results identify the exact rounds and record root they cover.
03 / ONCHAIN IDENTITY
Onchain anchoring
Loading…No chain claim is assumed from a local record.
Recording identity
Waiting for a recording.
———Recompute independently
The bundle includes a fixed prefix of committed records, their input frames and sampled telemetry, the initial model checkpoint, exact archived source, dependency pins and a replay command. The full public MaleCNS dataset is obtained separately and verified against the included checksum locks. It contains no wallet keys or private RPC credentials.
python -m venv .venv # Activate the environment and install requirements.txt. # Set STONKFLY_DATA to the checksum-verified MaleCNS dataset. python replay_recording.py --bundle .
Replay checks the record chain, artifacts, source and model identity, then recomputes every output, complete neural state and sampled time bin. The original Linux kernel is included when it matches the recorded build. A different environment may fail exact identity or replay checks; it must not be presented as an exact match.
Displayed trading account
The account panel reads public positions and fills for this address. The neural recording and commitments above retain their original run identity.
0x2bfe8fc49a666a44021a5273f60d3f94ec49255eArc CA / —