Roadmap

Where Mind Archive is going, and what is already finished. The detailed, authoritative version is MILESTONES.md.

The repository must remain runnable after every milestone.

Milestone 1 — Runnable foundation ✅

A clean, documented, runnable foundation with no product features yet.

React + TypeScript + Vite frontend with light and dark modes. Python + FastAPI backend with typed settings, health and configuration endpoints, and a small in-process event bus. Docker and Docker Compose. GitHub Actions CI. A GitHub Pages documentation foundation. Tests on both sides. Complete documentation and project memory.

Milestone 2 — ChatGPT importer ✅

The point at which Mind Archive becomes useful.

Import a ChatGPT export and store it as human-readable Markdown with JSON metadata. Establish the Importer adapter interface that every later provider will use. Treat imported files as untrusted throughout. A simple import screen that reports what happened in plain language.

Milestone 3 — Archive browser and search ✅

Browse and read imported conversations. Markdown rendering. A SQLite schema for metadata and indexing, with full-text search through SQLite FTS. The database stays rebuildable from the files on disk.

Milestone 3.5 — Import ergonomics ✅

Getting an export out of ChatGPT takes days, so importing one should take no effort at all. A watched inbox folder imports anything you drop into it. Re-importing reports what is genuinely new rather than counting everything again, and rewrites nothing that has not changed. Correct guidance about the export’s real timings. An optional faster route you run yourself in your own browser.

Milestone 4 — Organisation ✅

Tag your conversations, filter by tag, and combine that with search. Tags are stored in your archive files, not in the database, so they travel with it and survive anything happening to the index.

Projects were deliberately deferred — see decision D-026.

Milestone 5 — More providers, and export ✅

Import from Claude as well as ChatGPT, and download your whole archive as a zip of ordinary Markdown and JSON files that need nothing to read them.

The second importer did its real job: it exposed a bug in how exports were recognised that only a second provider could have found.

Milestone 6 — Optional cloud 🎯 next

Opt-in synchronisation to storage you choose. Disabled by default, never enabled silently. SyncProvider kept separate from StorageProvider. Candidate adapters — S3-compatible, WebDAV, self-hosted — added one at a time.

Milestone 7 — Production hardening and public release

Security review, packaging, a release process, and the public documentation site.

Not planned

A hosted service. An account system. A chat interface. Telemetry. Anything that requires a network connection for the core product to work.

Ideas that are not yet milestones are tracked separately in BACKLOG.md — nothing there is a commitment.