Port-owned 45/45 script-corpus proofs and strict 5k baselines.
Open research instrument · testnet4
Twelve separately implemented validators.
One immutable evidence boundary.
RosettaBitcoin is an artifact-backed experience report on verification infrastructure for agent-assisted Bitcoin consensus validators.
C++
port-owned proofC#
port-owned proofElixir
port-owned proofGo
port-owned proofJava
port-owned proofMojo
port-owned proofOCaml
port-owned proofPython
port-owned proofRust
port-owned proofSwift
port-owned proofTypeScript
port-owned proofZig
port-owned proofFrozen snapshot · 17 June 202612/12
corpus 45/45 · strict baseline 5kEach port emitted its own proof. Shared authorship and infrastructure mean these are not independent experimental replications.12ports with corpus + strict 5k
9canonical clean long-run lanes
1short maintenance artifact
0empty-state-to-tip proofs
01 / The experiment
What can the artifacts
actually support?
The snapshot records twelve separately implemented consensus validators. They execute their own validation decisions, but share one developer, repository, fixture corpus, blocker knowledge, byte source, evidence infrastructure, and some dependencies. They are not independent experimental replications.
02 / The substrate
Failures became
replayable evidence.
Raw blocks and runtime failures become provenance-rich blocker facts, reusable fixtures and rule cards, port-owned proofs, validating imports, and Project reports. Agreement is one check, not a substitute for negative testing or separate decision paths.
01Raw blocks + failures→
02Blocker facts→
03Fixtures + rule cards→
04Port-owned proofs→
05Project reports
03 / Snapshot outcomes
Milestones,
not completed nodes.
The immutable snapshot supports conformance and validation milestones. It does not contain an empty-state-to-tip proof, and no port passed the workspace’s binary full-node gate.
Canonical clean 50k, 100k, and post-100k evidence.
One short Java near-tip interval, not durable maintenance.
No empty-state-to-tip proof; Docker and live-node gaps remained.
04 / Secondary diagnostic
Pure Mojo,
bounded evidence.
A separate immutable supplement preserves the original pure-backend diagnostics without importing them into Project. The result is useful, but secondary to the snapshot study and explicitly outside comparable evidence.
Diagnostic supplement · DOI 10.5281/zenodo.22114337140,234
fresh-state to 100k, then resume to the recorded height
- Pure-Mojo secp256k1 · no native fallback
- 45 recorded shadow agreements
- Six must-reject classes · three mutation-killing modes
- Noncanonical · noncomparable · class-bounded
- Not the empty-state-to-tip binary gate
05 / Boundaries
Every result
has a boundary.
The evidence is most useful when its limits remain visible. These labels distinguish recorded milestones from the full-node result and controlled studies still waiting to be run.
Resume-to-tip
The Mojo diagnostic resumed from durable 100k state. It was not the empty-state binary gate.
Diagnostic crypto
The pure backend is a bounded correctness result, not a comparable performance entry.
Class-bounded negatives
Representative reject families are covered; completeness is not claimed.
Binary gate open
Empty state → tip → maintenance remains the workspace’s final end gate.
06 / The invitation
Reproduce the case.
Test the hypothesis.
Close the full-node gate, expand the negative corpus, run a controlled substrate ablation, or replicate the validators with an external team. Each new claim must arrive through an explicit evidence path.