Release Notes
This page tracks Quira's release history and planned milestones. Quira follows Semantic Versioning. During the alpha and beta phases, breaking changes may occur between minor versions.
v0.1.0-alpha (Planned Q3 2026)
Upcoming release
This release is currently in development. Features and timelines are subject to change.
The first public alpha release of Quira, intended for early adopters and contributors.
Core features
- Context Graph engine — Full browsing context capture with automatic topic clustering and semantic linking
- Context Spaces — Tab organization with GTD-inspired inbox, active, and archive states
- Local AI (Phi-3 Mini) — On-device inference for page summarization, natural language queries, and topic classification
- AI Privacy Levels — LOCAL mode (default) with opt-in CLOUD mode for heavier inference tasks
- Natural Language Query — Ask questions about your browsing history using the command palette (Ctrl+K / Cmd+K)
- Graph Visualization — WebGPU-accelerated interactive graph with force-directed layout, filtering, and timeline scrubbing
- Research Replay — Resume previous research sessions with full context restoration
- Export & Import — Export Context Graphs as JSON, Markdown, or Mermaid diagrams; import from Chrome/Firefox
Platform support
- macOS 13+ (Apple Silicon and Intel)
- Linux (Ubuntu 22.04+, Fedora 38+, Arch)
- Windows support planned for v0.2.0
Known limitations
- WebExtension compatibility is limited to a subset of MV2/MV3 APIs (see WebExtension API)
- Graph visualization performance degrades above approximately 10,000 nodes — pagination or time-based filtering recommended
- Cloud AI mode requires an API key from a supported provider (OpenAI, Anthropic, or local Ollama)
- No sync between devices in this release
Breaking changes
- None (initial release)
Foundation Build (Q1 2026)
The foundation build represents the internal development milestone that precedes the public alpha. This phase established the core architecture.
Completed milestones
| Milestone | Date | Description |
|---|---|---|
| Gecko fork baseline | January 2026 | Forked Firefox/Gecko engine, stripped telemetry, established custom build pipeline with just |
| Context Graph storage | January 2026 | SQLite + SQLCipher database schema with FTS5 full-text search and vector embeddings (sqlite-vss) |
| Context Spaces UI | February 2026 | Tab management UI with GTD-inspired workflow: Inbox, Active, Archive states |
| Local AI runtime | February 2026 | Integrated llama.cpp (CPU) and MLX (Apple Silicon GPU) for on-device inference |
| Basic graph visualization | March 2026 | Force-directed graph layout using WebGPU compute shaders with Canvas 2D fallback |
| Security layer | March 2026 | Network firewall, content filtering, permission system, and threat protection pipeline |
Architecture decisions
- Gecko over Chromium — Chosen for its more permissive licensing (MPL 2.0) and existing privacy infrastructure from Firefox
- SQLite over custom DB — Battle-tested, embeddable, with excellent full-text search (FTS5) and emerging vector search capabilities
- Local-first AI — Phi-3 Mini (4-bit quantized) as the default model, balancing capability with resource requirements
- Rust for core, TypeScript for UI — Rust handles the browser engine integration, graph engine, and AI runtime; TypeScript powers the UI layer
Versioning policy
Quira uses Semantic Versioning with the following conventions during the pre-1.0 phase:
- 0.x.0 — Minor version bumps may include breaking changes to the Plugin API and data schema
- 0.x.y — Patch releases contain only bug fixes and security patches, with no breaking changes
- Alpha — Public alpha releases are tagged with
-alphasuffix. Data schema may change between alphas (migration tools will be provided) - Beta — Beta releases stabilize the Plugin API and data schema. Breaking changes require a deprecation notice one release in advance
Alpha data migration
During the alpha phase, the Context Graph database schema may change between releases. Quira will attempt automatic migration, but we recommend exporting important research data before upgrading to a new alpha version.