What problem does it solve? Building a propose-then-review editing workflow — where humans or AI agents suggest changes that reviewers accept or reject — requires correctly wiring suggestion targets, capture pipelines, lifecycle events, and state application, which is easy to get wrong without canonical guidance. ## Core Features & Use Cases - Suggestion Targets & Mode: Tag DOM elements with data-velt-suggestion-target, register getters for complex multi-control values, and toggle suggestion mode with reactive state observation. - Three Capture Approaches: Auto-commit via onTargetEditCommit, deferred commit via the targetEditCommit event for validation gating, and manual startSuggestion/commitSuggestion for AI agent flows. - Lifecycle & Querying: Handle suggestionAccepted/suggestionRejected events idempotently, track statuses (pending/accepted/rejected/stale/apply_failed), detect drift, and query suggestions for custom UI badges and sidebars. - Use Case: A project management app where team members suggest changes to task fields (priority, assignee, due date) and the project owner accepts or rejects them from the Velt comment dialog, with accepted values applied to app state. ## Quick Start Ask the agent to add Velt suggestion mode to your React form inputs so edits become proposals that reviewers can accept or reject.
