TL;DR
Threlmark’s published architecture describes a local-first project management app that treats files on disk as the system of record. The design uses plain JSON, atomic writes and one file per card to support offline use, tool interoperability and agent workflows, though real-world sync limits remain unclear.
Threlmark, a Next.js and TypeScript project management app, is being presented as a local-first system that uses plain JSON files on disk as its source of truth rather than a traditional database, a design choice that could affect offline work, portability and AI-agent integration.
The report from Thorsten Meyer AI says Threlmark stores project data under a default local root, ~/.threlmark, with separate files for the manifest, dependency links, project metadata, board ordering and individual work items. Each card is stored as its own JSON file, while folders also hold suggestions, handoffs, reports and human-readable roadmap files.
According to the source material, the central design decision is that the on-disk layout acts as the API. The user interface and external tools are expected to read and write the same file structure, using the same rules. The report says this makes data inspectable with common tools such as cat, grep, diff, git and standard backup utilities.
The article says Threlmark uses two main safety patterns to make file-based state workable: atomic writes and one file per item. Atomic writes mean data is first written to a temporary file in the same directory, synced, and then renamed over the target. The report says this should leave either the old complete file or the new complete file after a crash, rather than a partial write.
Why It Matters
The design matters because it challenges a common assumption in project management software: that the server or database must be the system of record. If the architecture works as described, users could keep their project state in a form that is easier to inspect, back up, version, sync or move between tools.
For developers and teams using AI agents, the architecture also points to a more direct workflow. The report describes agents writing reports through REST or by dropping files into a reports folder. A completed report can move a card to Done without requiring a user to drag it manually, according to the source material.
The practical impact will depend on how well Threlmark handles concurrent edits, malformed files, sync conflicts and larger workloads. The article presents a disciplined file model, but it does not provide independent benchmarks, user adoption figures or production incident data.
Python in Action: 60 Mini Projects to Automate Everything (Part 1): Practical CLI Tools, File Automation, and Data Cleaning with CSV, Excel, and JSON
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Background
Threlmark is described in the source material as a Next.js app built on TypeScript and JSON-on-disk, with no database, cloud account or remote server of record. The report frames this as part 2 of a Threlmark series and says the project is MIT-licensed.
The architecture uses a project folder structure that separates card state from board ordering. Instead of putting every card into one large roadmap file, each item gets its own JSON file. Lane order lives in board.json, and the board reconciles state on read, according to the article.
The report also says Threlmark avoids storing values that can be computed from item state. Priority, stale status, cycle time, throughput and work-in-progress counts are described as derived values, reducing the chance that displayed metrics drift from the underlying files.
“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
“Just use files is easy to get wrong.”
— Thorsten Meyer AI report
“A handoff is a first-class flow event.”
— Thorsten Meyer AI report
Information Technology Project Management (Mindtap Course List)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
What Remains Unclear
Several details remain unclear from the provided source material. The report does not state how many users or teams are using Threlmark, whether the architecture has been tested under heavy concurrent access, or how it resolves conflicts created by external sync tools such as Dropbox or git.
It is also unclear what parts of the system are currently released, experimental or planned. The source describes REST reporting, filesystem fallback and automatic card movement, but does not include external verification or release notes for each feature.
version control for JSON files
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
What’s Next
The next milestone is whether Threlmark’s file contract proves reliable outside the described architecture. Readers should watch for code releases, documentation of the disk schema, conflict-handling guidance, real-world usage reports and follow-up posts in the Threlmark series.
Blue Summit Supplies 50 Project File Folders, Reinforced Straight Cut Tab, 2 Inch Expansion Pocket, Preprinted Project Management Folder Design, Letter Size, Manila, 50 Pack
Expandable Project File Jackets with Preprinted Project Tracking Layout – Dedicated sections for project details, task tracking, follow-up…
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What is the actual development?
Thorsten Meyer AI has published a report describing Threlmark’s architecture, which treats local JSON files on disk as the source of truth for a project management app.
Does Threlmark use a database?
According to the source material, Threlmark does not use a traditional database or cloud server as the record. It stores project data in plain files under a local data root.
How does the system reduce file-write risk?
The report says Threlmark uses atomic writes and stores each work item in its own file. That approach is meant to reduce partial writes and limit write conflicts between tools.
Why does this matter for AI agents?
The architecture gives agents a file-based way to receive handoffs and report results. The report says a completed agent report can move a card to Done through REST or a filesystem fallback.
What remains unconfirmed?
The source does not confirm adoption, production reliability, large-scale performance or conflict behavior under external sync tools. Those areas remain open pending more evidence.
Source: Thorsten Meyer AI
