Labello documentation

See the product glossary for canonical UI actions and concepts.

Start with getting started, then choose the guide for your work. These pages describe the implemented application. Code and tests resolve any disagreement with the prose.

Use Labello

Guide	Contents
Getting started	Local setup, first dataset, sign-in, connection troubleshooting
Annotation and review	Boxes, skeletons, controls, corrections, navigation, drafts
Stylus input	Pen controls, browser/device matrix, emulated checks, physical iPad procedure
Dataset administration	Schema reuse, roles, image ingestion, inspection, statistics, snapshots
Model prelabels	ONNX model supply, inference, selection, batch hints, reset and provenance
Dataset import	Supported YOLO/COCO profiles, source upload, planning, coverage, publication
Guided migration	Box-to-skeleton work, exclusions, discoveries, companion reconciliation
Dataset export	Selection, completeness, artifacts, preservation and round trips
Assignment	Eligibility, leases, completion, balance, previous review, correction transactions
Contribution scoring	Rewards, deductions, daily tiers, focus workflows, leaderboard meaning
Current limitations	Partial features, unsupported behavior, operating constraints

Operate a server

Guide	Contents
Configuration	Server and browser settings, authentication, environment, resource limits
Release and deployment	Release verification, guest transaction, readiness, rollback
Guest setup	Debian LXC provisioning, permissions, user services, Caddy
Operations	Logging, health, capacity, backup, restore, upgrades, incident handling
Persistence and recovery	Artifact authority, event transactions, schema compatibility, recovery
Event history	Workflow event replay, historical compatibility, server-owned commands

Develop Labello

Guide	Contents
Contributing	Setup and change/review workflow
Architecture	Crates, dependency direction, shared interfaces
Workflow policy	Domain validation, API authorization, storage transaction boundaries
HTTP API	Routes, access, wire types, limits, responses
UI design	Layout, interaction, accessibility, viewport acceptance
UI implementation	State, async requests, rendering, browser recovery
Native inspector	Headless setup, presets, live inspection, evidence
Verification	Canonical checks, risk profiles, CI, review gates
Parallel development	Worktrees, dependent branches, combined verification
Documentation and wiki	Editing, checking links, publication, archive boundary

Update the relevant current page when behavior changes. Keep each detailed contract in one place and link to it from user guides. Documents have no owner, status, date, or revision headers; Git records their history.

Plans, proposals, target requirements, and delivery records belong in the repository-only archive. They are preserved for context and excluded from the wiki. Planned work is tracked in GitHub issues.