Autonomous engine, gallery, and capture trust
Current public download. OWOJI now points toward an autonomous agent engine: users give a goal, agents run, outputs ship, and templates compound over time.
- Added Gallery, Goal, Skills, Templates, Metrics, Changelog, and Feedback web surfaces.
- Added backend contracts for goals, agent runs, outputs, templates, audit events, feedback, and metrics.
- Updated the public story around Goal → Agents Run → Outputs Ship → Templates Compound.
- Added an in-app Update pill and a manual Check for Updates action.
- Notifications now surface update status, capture state, and setup reminders.
- Added a clearer Start New Session path after capture review and packaging.
- Sessions can be opened again so users can continue from a completed capture.
- Manual detail insertion is easier to find between workflow events.
- Capture Signals settings let users choose app switches, browser tabs, prompt focus, video context, screen ratio, window count, tab count, keyboard source, and aggregate typing signals.
- Keyboard tracking is opt-in and aggregate-only; OWOJI does not store passwords or raw keystroke text.
- Head movement and eye tracking are listed as future signals and remain disabled until explicit camera/gaze permission work ships.