TL;DR
Thorsten Meyer AI has published a technical report on Threlmark, a Next.js and TypeScript project tool that treats local JSON files as the system record. The report says the design is meant to make project data inspectable, portable and usable across tools, while open questions remain around real-world conflict handling and broader adoption.
Thorsten Meyer AI has published a technical report on Threlmark, a Next.js project-management app that uses plain JSON files on a user’s local disk as its record instead of a traditional database, cloud server or account system, a design choice that could affect how teams think about portability, offline work and AI-agent workflows.
The report says Threlmark’s data root defaults to ~/.threlmark, with project metadata, board state, item files, suggestions, handoffs, reports and a human-readable roadmap stored directly on disk. According to the source material, the project’s central decision is that “the on-disk layout IS the API”, meaning the file structure is treated as the stable interface for the app and for outside tools.
Thorsten Meyer AI describes Threlmark as a Next.js and TypeScript app built on JSON-on-disk storage under an MIT license. The report says the app avoids a server-of-record: the files themselves are the record, and both the user interface and external tools interact with those files through the same rules.
The architecture relies on two file-safety patterns, according to the report: atomic writes and one file per item. Atomic writes use a temporary file in the same directory, followed by rename() over the target, so a crash should leave either the prior complete file or the new complete file. The one-file-per-item model is intended to reduce collisions when multiple tools write data, while lane ordering is stored separately in board.json and reconciled when Threlmark reads the board.
Why It Matters
The report matters because it frames local files as more than a storage detail. If the local-first approach works as described, users could inspect, back up, diff, sync or move project data with ordinary file tools instead of relying on a hosted service or proprietary database export.
That design also has consequences for AI-assisted software work. The source material says Threlmark treats agent handoffs and reports as first-class workflow events. An agent can receive a brief, report back through a REST endpoint or by writing a report file, and a completed report can move a card to Done. That matters for teams experimenting with AI agents because it creates a file-based path for work status to return to the board even if the server is unavailable at finish time.
The trade-off is that Threlmark’s model shifts responsibility to the file contract. Data safety, compatibility and outside-tool behavior depend on the discipline of the file layout, write patterns and reconciliation rules. The report argues that this makes the system easier to inspect and extend, but the source material does not provide independent production benchmarks or user adoption figures.
As an affiliate, we earn on qualifying purchases.
Background
Local-first software aims to keep user data available on the user’s device and reduce dependence on continuous network access. Threlmark’s version of that idea is narrower and more concrete: project data lives in a home-directory folder as JSON files, and the app treats those files as the shared hub for the interface and other tools.
The report lists several consequences of the design. Each artifact can be viewed or compared with common command-line and version-control tools. Backups and sync can use normal file-copy, Dropbox or Git workflows. Tools written in other languages can participate by reading and writing the same files, provided they follow the contract.
The report also says derived values are computed rather than stored. Priority, stale-work markers, cycle time, throughput and work-in-progress counts are calculated from item state or lane data so displayed metrics do not drift from the underlying JSON files. Priority is described as a weighted formula based on impact, evidence, fit and effort.
“The on-disk layout IS the API.”
— Thorsten Meyer AI report
“There is no server-of-record — the files are the record.”
— Thorsten Meyer AI report
“Every artifact is a file you can cat, diff, grep, commit.”
— Thorsten Meyer AI report
“Unknown keys are preserved, so the contract is forward-compatible.”
— Thorsten Meyer AI report
Data Recovery Stick for Windows Data Recovery Software – Photos, Files
- Easy to Use: Plug and recover files automatically
- Wide Compatibility: Supports Windows Vista to 11
- File Type Support: Recovers photos, documents, music, PDFs, and more
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
What Remains Unclear
Several points remain unclear from the supplied source material. The report does not state how many users or projects currently rely on Threlmark, whether the architecture has been tested under heavy concurrent use, or how it handles every class of conflict across multiple syncing services.
It is also not clear whether the file contract is documented in a formal public specification, how schema changes are versioned over time, or what guardrails exist for third-party tools that write malformed JSON or partial data outside the prescribed atomic-write pattern.
REXBETI 25Pcs Metal File Set, Tool Sharpening, Woodworking, Finishing File
- Material: T12 Drop Forged Alloy Steel
- Number of Files: 25-piece set with various files
- Included Tools: Sander paper, gloves, wire brush, case
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
What’s Next
The next questions are whether Threlmark’s file contract becomes a stable public interface for outside tools, whether the project publishes more detail on schema versioning and conflict recovery, and whether real-world users adopt the local-first model for AI-assisted project work. Future reports in the series may clarify deployment details, portfolio scoring and how the architecture behaves when several tools write to the same workspace.
As an affiliate, we earn on qualifying purchases.
Key Questions
What is the actual news development?
Thorsten Meyer AI published a technical report on Threlmark’s local-first architecture, describing how the app uses a user’s local disk and plain JSON files as the system record.
What is confirmed by the source material?
The source states that Threlmark is a Next.js and TypeScript app, stores data under ~/.threlmark by default, uses JSON files rather than a database, and relies on atomic writes plus one file per item for safety.
What is claimed but not independently verified here?
The report claims the architecture improves inspectability, portability, offline use and tool interoperability. Those claims are attributed to the source material; the supplied material does not include independent benchmarks or third-party user data.
Why use one file per card?
According to the report, one file per item reduces write collisions because an outside tool can create or update an item file without editing the board-order file. Threlmark then reconciles the board when it reads the project.
What happens next?
Readers should watch for more detail on the file contract, schema versioning, conflict handling and adoption by external tools or AI-agent workflows.
Source: Thorsten Meyer AI
