Rust TUI GTD Todo (OmniFocus-inspired)

Contents

• Concept
• Features
• Current Status

Concept

Keyboard-driven Rust TUI task manager built around a GTD workflow. Stores tasks in a local SQLite database with support for projects, hierarchical tags, and availability-based next-action computation. Optimized for rapid inbox processing — single-key field mode for triage, project/tag assignment, and batch operations with sub-second interactions at 10,000+ actions.

Includes a one-time importer for OmniFocus 4 .ofocus-package files, mapping projects, folders, tasks, tags, notes, and dates into the local database.

Features

• GTD views: Inbox, Projects, Tags/Contexts, Next, Waiting, Someday/Maybe, Review
• Single-key field mode for fast inbox triage with overlay selectors
• Hierarchical tags with descendant-inclusive filtering
• Next actions computed from availability rules (active, not completed, defer date, not blocked)
• Per-project review intervals with review screen
• OmniFocus 4 one-time import (read-only, .ofocus-package)
• Architecture hooks for future local-only, opt-in LLM processing suggestions

Quick Facts

What This Is

A fast, keyboard-driven Rust text UI (TUI) task manager inspired by OmniFocus, built around a GTD workflow. It stores data in an owned SQLite database, supports projects and hierarchical contexts/tags, and is optimized for rapid inbox processing (triage + tagging) with minimal friction.

An OmniFocus 4 .ofocus-package importer provides one-time migration into the local database so the tool can replace OmniFocus for day-to-day use.

Core Value

Process an inbox full of captures into correctly-organized next actions (project + tags + defer/due) at high speed, with sub-second interactions.

Requirements

# Validated

(None yet  ship to validate)

# Active

• Keyboard-first TUI with OmniFocus-like navigation: Inbox, Projects, Tags/Contexts, Next, Waiting, Someday/Maybe, Review
• Fast inbox processing sprint: single-key field mode with overlays for project/tag/date assignment and batch operations
• Owned SQLite data model supporting projects + actions + hierarchical tags, with filtering that includes descendants by default
• Next actions computed via availability rules (active + not completed + defer<=today + not blocked), with performant list queries
• OmniFocus 4 one-time import from .ofocus-package mapping projects/folders, tasks, tags, notes, and core dates into SQLite
• Per-project review interval and review screen showing projects due for review
• Foundation for future local-only, opt-in LLM suggestions (initial target: inbox processing suggestions)
• Claude Code integration for batch inbox processing: export inbox items, auto-suggest projects/tags/contexts with confidence levels, thematic grouping, and batch acceptance UI

# Out of Scope

• Ongoing sync with OmniFocus  one-time import only for v1
• Web/mobile UI  terminal-only for v1
• LLM features shipping in v1  only architecture hooks and data needed for later

Context

• Motivation: OmniFocus is powerful but painful to work with; the goal is a replacement that keeps the familiar mental model while prioritizing speed and ergonomics.
• Platform: macOS-only initially.
• Scale expectation: up to ~10k actions.
• UX emphasis: vim-like keyboard navigation; single-key field mode for triage; tag selection must be extremely fast.

Constraints

• Performance: Sub-second UI response; queries and filtering must remain snappy at ~10k actions.
• Tech stack: Rust TUI application with SQLite persistence.
• Import: OmniFocus 4 .ofocus-package supported; importer is read-only and one-time.
• Privacy: Future LLM integration must be local-only and opt-in by default.

Key Decisions

Last updated: 2026-02-20 after initialization

Current Status

2026-02-25 - Completed 13-01-PLAN.md