TL;DR
Thorsten Meyer AI has published a technical report on Threlmark, a local-first project tool built as a Next.js app over JSON files stored on disk. The report says the file layout is treated as the API, allowing external tools and AI agents to read, write and complete work items without a database or cloud account.
Thorsten Meyer AI has detailed the architecture behind Threlmark, a local-first project management tool that runs as a Next.js app and stores its working state in plain JSON files rather than a database, cloud service or user accounts. The report matters because it presents file storage not as an export format, but as the primary contract for the app, outside tools and AI coding agents.
The report says Threlmark’s data root defaults to ~/.threlmark, with a manifest, dependency graph, project folders, board state, one JSON file per item, suggestion drop zones, handoff records, agent reports and a human-readable roadmap file. According to the source material, every artifact is meant to be readable with standard file tools and portable through ordinary backup, sync or version-control workflows.
The core design choice is that the on-disk layout is the API. In the report’s wording, “There is no server-of-record — the files are the record.” That means the user interface and external tools are expected to follow the same file discipline instead of depending on a central service.
Threlmark uses atomic writes and one file per work item to reduce common file-state risks, according to the report. It writes to a temporary file in the same directory, fsyncs it, and then renames it over the target. The report says lane order is held in board.json and reconciled on read, while item data lives in separate files so external tools can add cards without editing the board file directly.
Why It Matters
The design is aimed at developers and AI-assisted builders who want project state to remain inspectable, portable and tool-accessible. If the system works as described, a user could grep task data, commit it to Git, back it up with normal filesystem tools, or let another program participate by reading and writing JSON.
The report also ties the architecture to agent workflows. Threlmark treats an AI handoff as a first-class project event: a card can be assigned to an agent, the agent can return a report through a REST endpoint or by dropping a JSON report file, and a completed report can move the card to Done.
That approach addresses a current pain point in AI-assisted software work: tracking whether delegated tasks actually shipped. The report frames Threlmark less as a column-based board and more as a way to rank work across projects and close the loop when agents finish assigned tasks.
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
The source describes this as part 2 of a Threlmark series and labels the project with Next.js, TypeScript, JSON-on-disk and MIT. It positions Threlmark as related to an earlier tool whose priority formula is reused, so imported cards rank the same way.
Several values are derived rather than stored, according to the report. Priority is calculated from impact, evidence, fit and effort; age, cycle time, throughput and WIP counts are computed from item state and append-only movement history. The stated goal is to keep displayed metrics aligned with the files on disk.
The report also describes a portfolio view that ranks globally addressable items by a status-weighted score, giving more weight to development work and some extra lift to blocked items. Deployment details are only partly visible in the source material, including a static read-only demo path and a personal Node path that appears truncated.
“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
“A handoff is a first-class flow event.”
— Thorsten Meyer AI report
Real-World Android App Projects with Kotlin and Jetpack Compose: Build Production-Style Android Apps with Modern Architecture, API Integration, State Management, Local Data Storage, Practical Projects
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
What Remains Unclear
The source material does not state a release date, current user adoption, production-readiness status, security model, or how Threlmark handles conflicting writes across synced devices. It also does not confirm how well the design performs with very large boards or multi-user teams.
Some deployment information in the supplied material is incomplete, including the description of the personal Node path. Claims about portability, interoperability and restartability come from the project report and have not been independently verified here.
15.6 Inch Digital Calendar Wall Touch Screen Chore Chart, 32GB Electronic Calendar + Smart Family Planner, Wall and Desk Mountable, Gifts for Women Men, Gifts for Mom Dad
📆 𝑺𝑴𝑨𝑹𝑻 𝑪𝑨𝑳𝑬𝑵𝑫𝑨𝑹: 𝑬𝑨𝑺𝒀 𝑺𝑬𝑻𝑼𝑷 & 𝑺𝑴𝑨𝑹𝑻 𝑺𝒀𝑵𝑪 📆 Sync your calendar across platforms in minutes—compatible with Google,…
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
What’s Next
The next questions are practical ones: whether the project publishes stable file-format rules, how it handles sync conflicts, and whether external tools begin to write against the ~/.threlmark layout. Readers following the project should watch for release notes, schema documentation, setup instructions and examples of agent reports moving cards to Done in real workflows.
AI-Assisted Coding: A Practical Guide to Boosting Software Development with ChatGPT, GitHub Copilot, Ollama, Aider, and Beyond (Rheinwerk Computing)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What is the actual news development?
Thorsten Meyer AI published a technical report explaining Threlmark’s local-first architecture and its decision to make the on-disk JSON layout the main contract for the app and external tools.
What is confirmed by the source material?
The report says Threlmark is a Next.js and TypeScript app that stores project data in plain JSON files under ~/.threlmark, uses one file per card, applies atomic writes, derives metrics on read, and supports AI agent reports through REST or filesystem drop zones.
What is claimed but not independently verified here?
The claims that the design is portable, interoperable, restartable and safe enough for file-based state come from the report. The supplied source does not include outside testing, benchmarks or user adoption data.
Why does this matter to readers?
For developers, it offers a project-management model where data remains inspectable and scriptable. For teams using AI coding agents, it proposes a concrete way to track handoffs and mark work complete when an agent reports that it shipped.
What remains unclear?
Release timing, conflict handling across sync tools, large-board performance, security boundaries and production deployment details are not fully established in the provided source material.
Source: Thorsten Meyer AI
