Cryptographic Evidence · Solidity · Foundry · Research
An open source laboratory documenting the journey of building provable off-chain and on-chain systems.
bunje.io is a personal lab for blockchain software engineering. Everything is built openly, documented publicly, and published on GitHub.
// Personal research. Does not reflect the position of my employer. No finished product. Learning by building. The GitHub repositories document the current state: early, honest, open.
Published in Eisenbahntechnische Rundschau (ETR), an established technical journal for infrastructure and transport.
Examination of cryptographic evidence mechanisms for regulated approval and infrastructure procedures.
Klaus Walter: Blockchain in der Planfeststellung: Möglichkeiten für Effizienz und Nachvollziehbarkeit. In: Eisenbahntechnische Rundschau (ETR) 5/2026, DVV Media Group.
Journal at publisher →Architecture decisions, security considerations, and design tradeoffs will be published here as the projects evolve. Articles on state machine design, cryptographic evidence patterns, and lessons from applying blockchain to regulated workflows.
The stack is deliberately narrow: Solidity and Foundry until the fundamentals are solid, then Python for the verification layer.
Secure Solidity contracts, starting with access control, state machines, and workflow logic for regulated processes.
Testing, deployment, and debugging with Foundry. Every contract has tests before it is deployed.
web3.py for the off-chain verification layer: reading and checking hashes against the on-chain anchor, not just writing them.
Everything above is developed in public on GitHub. Code, decisions, and progress are visible from the first commit; there is no proprietary lock-in.
All projects are open source. Status reflects where each project currently stands — not where it is marketed to be.
Forsblock is the research protocol: cryptographic evidence for regulated approval and infrastructure workflows, modelled as state machines on-chain. ClaimWorkflow is the first application built on it, a workflow for claim and approval processes with a smart contract backbone, deployed on Sepolia testnet.
Two side tracks. One explores Real World Asset tokenisation and the cryptographic representation of regulated assets on-chain, where RWA meets infrastructure practice. The other is a smart contract protocol for claim and site workflow management, a state machine for structured, verifiable approval sequences.
Protocol design experiments and architectural explorations. Where ideas are tested before they become contracts. Public from the first commit.
Planned web3.py tooling for the off-chain verification layer. Hashing documents, checking them against on-chain anchors. No code yet. The problem is clear, the timing is not.
This is where the learning actually stands. No retrospective polish — just the sequence as it happened.
The repositories are the product. Not polished demos — real code, real tests, real decisions. Including the dead ends.
Code is the documentation. Each repository includes Foundry tests, deployment scripts, and notes on design decisions. The commit history is honest — including the refactors and the mistakes.
An experimental protocol exploring cryptographic evidence mechanisms for regulated workflows. A research project built in public, without a product or a company behind it.
SHA-256 hashing, digital signatures, qualified timestamps, immutable audit trail. Everything else is built on this layer.
Public blockchain anchoring. Tamper-resistant, long-term verifiable, vendor-independent. Anchors the hashes from Layer 01.
Construction law translated into machine-verifiable state machines. Roles, workflow gates, variation order protocols, coded rather than described. Intended as an open standard.
First application built on Forsblock. Cryptographic state machine for claim and approval workflows. Deployed on testnet, experimental.
The ClaimWorkflow state machine runs through five phases:
Submission, hash, anchoring
Inspection, signature, timestamp
Bilateral, deadlines, protocol
Decision, both signatures
Export, auditable
// Construction law translated into machine-verifiable process logic · Forsblock Protocol · Klaus Walter / bunje · 2025–2026 · Experimental
It's a research project, built in public because the problem is real and worth studying openly — not a product with customers or a token behind it. Anyone can read the code, challenge the design, or build on the concept.
The evidence layer is not an alternative to AI. It is the precondition that makes AI usable in regulated procedures at all: the record has to be verifiable before any inference on top of it is worth anything.
Point an agent at a project folder and it finds seventeen versions, three file formats, and no record of which one was approved. It will still answer. That is the danger: the output reads confidently and cannot be checked.
Hashes, timestamps and signatures turn a document into a checkable claim: this file, this state, this approver, this moment. An agent working against anchored data can cite evidence instead of producing prose.
Deadline monitoring, version conflicts, missing approvals, a Z0 in the Bautagebuch against a Z2 on the Lieferschein. Mechanical checks against a verifiable record. The administrative decision stays with the human, and that is the legal order, not a limitation.
There is no agent code in any repository. This section describes a direction, not a feature. The off-chain verification layer (Python, planned 2027) comes first. Agents come after, if at all.
The protocol produces verifiable records; whether an agent later reads them is a separate, open question with its own problems — liability, explainability, and the boundary between preparing a decision and making one. Building the evidence layer first is the whole point. An agent on unverifiable data reproduces the problem faster, it does not solve it.
Not a startup founder. Not a crypto native. An infrastructure engineer exploring how cryptographic software can improve evidence, transparency, and workflow integrity.
// Personal research. Does not reflect the position of my employer. No finished product. Learning by building. The GitHub repositories show the current state: early, honest, open.
Infrastructure practice meets cryptographic software engineering. If any of this resonates — write.
"Mein Gott, Walter…" — That is what colleagues said when I named a situation everyone knew but no one had articulated precisely. That is still the starting point of everything here.