From Cashu Proofs To OpenETR¶
Policy Proposition¶
Digital continuity requires more than keeping an application running. People need to retain usable records when software changes, recover them when services fail, and evaluate their significance without depending exclusively on the system that first presented them.
The development path from lnfastapi through Safebox and the Mainstay family to OpenETR illustrates how those requirements can emerge from practical work. A wallet-storage insight became a private-record architecture, then a way to preserve files, and ultimately a framework for deriving consequential state from independently verifiable evidence.
The application can help people use their records without being the permanent home of their data or the exclusive authority for what those records mean.
This brief is a project retrospective and an account of the policy lessons drawn from it. It is not a claim that the underlying protocols were invented by these projects, or that cryptography eliminates operational and institutional responsibilities.
The Starting Point: A Payment Application¶
The local Git history of lnfastapi begins on June 26, 2022. The safebox-2 repository begins on July 23, 2024. These are repository milestones, not claims about the first conception or public release of every capability that followed.
The work began with a concrete problem space: making payments and wallet functions usable. It did not begin with a complete theory of portable digital evidence. That distinction matters. The broader architecture emerged through building, using, breaking, and improving working systems.
September 2024: Proofs Could Be Events¶
In the project creator's recollection, the decisive insight came in September 2024: Cashu proofs could be stored as Nostr events on relays. The immediate follow-up question was, “why not records too?”
The insight was about separating the wallet's application interface from the place where its private data was preserved. Sensitive proof material could be protected in encrypted event content, while relay infrastructure supplied a means of publishing and retrieving that content.
Storing a proof does not establish that it remains spendable. Mint checks, recovery procedures, and coordination between writers still matter. Relay storage is not a substitute for payment finalization or a mechanism that independently prevents double spending.
Nevertheless, the architectural opening was substantial: if this approach could preserve sensitive wallet material, it could also preserve private notes, documents, credentials, and other user records. The wallet could become a general-purpose private record environment without making every record a financial instrument.
From Records To Files Through Blossom¶
The next expansion was from event-carried records to larger files through Blossom. Events could retain record metadata and references; content-addressed storage could hold the associated bytes. Private attachments could be encrypted, with the information needed to access them protected inside private records.
This separated several responsibilities that applications often combine:
| Responsibility | Architectural role |
|---|---|
| Attribute an event | A signature that can be checked against its signing key |
| Protect private content | Encryption and controlled access to decryption material |
| Identify exact file bytes | A cryptographic digest |
| Preserve and retrieve material | Relay and file-storage services, with appropriate replication and access arrangements |
| Present useful workflows | Applications that interpret and display the records |
A digest permits a downloaded file to be checked independently of the server that supplied it. It does not guarantee that a copy is available. Encryption protects content, but does not make key loss harmless or erase all metadata exposure. These boundaries became practical design requirements rather than reasons to abandon portability.
A Family Of Components, Not One Indispensable Application¶
As the work developed, responsibilities became more explicit across the Mainstay family. The Mainstay product vision describes a local-first entry point coordinating independently useful components: Safebox for user workflows, Acorn for portable keys, funds, records, signing, and recovery, Spurline for event continuity, Grove for blob storage, and Clear for bounded local currencies and vouchers. Stroma supplies a narrower runtime boundary for Nostr operations and, more recently, pooled Blossom retrieval.
The ambition is continuity across connected and disrupted conditions. It should not be read as a promise that every payment can finalize offline, that all components have identical maturity, or that infrastructure ceases to matter. Confirmed funds and pending value must remain distinguishable; local access and external settlement are different capabilities.
The separate safebox-acorn repository begins on July 27, 2026. OpenETR's repository begins earlier that year, on April 25. The projects therefore developed in parallel and informed one another. The conceptual sequence in this retrospective is not a claim that each repository was created only after the previous project was complete.
OpenETR: From Preserving Records To Deriving Consequences¶
The storage architecture answered an increasingly general question: how can people preserve and recover private material without tying its lifetime to one application?
OpenETR addresses a further question: what can another party establish from the evidence, and what state follows under defined rules?
Its axioms make the distinctions explicit:
- A digest identifies a Digital Artifact independently of its location.
- A signature attributes a statement to a signing key; it does not alone establish that key's institutional authority.
- An anchor begins a candidate Evidence Graph, not an automatically recognized or universally unique original.
- Linked evidence, validated and evaluated under defined rules, supports derived Consequential State.
- Recognition and effect remain questions for the relevant relying party, institution, agreement, or law.
This was more than adding a verification badge to a stored file. It made the basis for a conclusion explicit and potentially reproducible by another implementation. The application could show its evidence and rules rather than ask a user to accept its displayed status as the sole source of truth.
The distinction between wallet records and OpenETR evidence is important. Mutable private application state can be maintained through replaceable relay events. OpenETR's consequential history is derived from linked signed evidence; it is not simply the latest private database snapshot given a new name. Nor does every private Safebox record automatically become an OpenETR Evidence Graph.
The resulting policy proposition is described in Why OpenETR: make consequences portable while keeping verification separate from recognition. Shared evidence need not force every participant to adopt the same recognition policy, but reproducible protocol conclusions require compatible rules and sufficient evidence.
A Barcode Makes The Principle Concrete¶
The GS1 Digital Link work provides a visible expression of this evolution. An ordinary camera can open a website from a barcode. An independent reader can instead extract identifiers and a supplied digest, seek supporting evidence through its own configured sources, and evaluate the result without treating the printed website as authoritative.
The URL is then an acquisition input, not an instruction to trust its host. A digest checks correspondence with exact bytes; a signed anchor can associate those bytes with product identifiers. Recognition of the issuer, the truth of the product claim, and the association with the physical item remain separate questions. A substituted barcode is not made trustworthy merely because its references and signatures are internally consistent.
See GS1 Digital Link And Verifiable Product Information for the detailed model and its limits. The important continuity with the original wallet insight is the reduction of dependence on one application or hosting location—not the elimination of all trust.
Migration As A Practical Test Of Continuity¶
A recent record migration supplied a small but meaningful test. In the creator's reported run, 38 eligible private records were accounted for: ten were already present at the destination and 28 were subsequently copied and verified. Earlier attempts exposed missing source locations and unavailable file endpoints; supplying current and older Blossom sources enabled recovery.
The operation decrypted and checked attachments, encrypted them with fresh keys for the destination, and verified readback. It left source records unchanged and excluded funds and operational payment state. That demonstrated a bounded form of continuity across software generations and private-key migration—not a complete backup guarantee or proof that every historical record was found.
The later BlossomPool integration improves acquisition by querying candidate servers concurrently and accepting a digest-verified response. It does not solve competing wallet writers, repair uncertain payment state, or change the meaning of the recovered records.
This experience also qualifies the portability claim: formats, legacy lookup conventions, available copies, and migration tools all matter. Portability is something to exercise and verify, not merely declare in an architecture diagram.
Policy Lessons¶
Treat Exit And Recovery As Normal Capabilities¶
Procurement and deployment decisions should ask whether users can recover their material and move it to another implementation or identity. An export button is insufficient if the result lacks attachments, usable decryption material, provenance, or a documented interpretation.
Separate Integrity From Availability And Recognition¶
Digest checks, signature validation, replication, and institutional recognition answer different questions. A resilient design states which question each component answers and reports uncertainty where the evidence is incomplete.
Preserve Evidence Alongside Conclusions¶
A displayed status is useful, but it should not be the only surviving record of how a consequential decision was reached. Retaining relevant signed evidence, ruleset versions, and evaluation context makes later review and independent verification possible, subject to access and preservation requirements.
Make Local Continuity Honest¶
Local infrastructure can preserve access and pending work during disruption. It must not represent unavailable external confirmation as completed settlement. The same discipline applies to evidence: failure to retrieve a notice does not prove that no such notice exists anywhere.
Keep Cryptographic Capability Distinct From Authority¶
Possessing a private key enables cryptographic operations. It does not, on its own, establish a person's identity, an organization's mandate, or the legal effect of an action. This boundary becomes more important as software agents participate in payment and record workflows.
The Continuing Research Direction¶
The proposed “Synthetic Minds” and “Simple Mind” work extends the same question to agents: could goals, context, and durable state persist outside a particular LLM or execution host, with models serving as replaceable reasoning services? That is a research direction, not a demonstrated consequence of using relays, and it does not establish consciousness, secure delegation, or trace-free host execution.
OpenETR offers a complementary question for such experiments: which actions create consequential evidence, under what authority, and how can another participant verify what follows? The agent authority brief explores that boundary.
Retrospective Conclusion¶
The defining September 2024 question was modest: if Cashu proofs can be stored as events, why not records too? Blossom extended the answer to files. The Mainstay family gave the responsibilities distinct components and a continuity vision. OpenETR brought the reasoning into a framework for evidence, rules, consequential state, and recognition.
The enduring idea is not that every system should become a wallet, nor that every record should become a transferable instrument. It is that people should not have to lose their data, its evidentiary value, or the ability to evaluate it independently when an application or service changes.
Basis And Scope Of This Retrospective¶
This account was prepared on October 11, 2026 from the project creator's recollections and reported migration results, local repository history, and the project documentation linked above. September 2024 is a recalled insight, not a date independently established by a particular commit. Repository start dates refer to root commits in the inspected histories. The migration is a reported operational example, not an independent audit.
The brief interprets one development lineage. It makes no claim of priority over Cashu, Nostr, Blossom, or other work on cryptographic records, and makes no independent legal-compliance claim.