A review tool where human feedback meets AI action. Anchored comments on plans, designs, and prototypes — structured JSON exports that agents consume, act on, and resolve.
Three content types, six anchor types, one clean API. Revi keeps humans and agents in sync.
Markdown documents with heading anchors, text selection comments, inline search, and collapsible document index.
Images on a dark canvas with numbered annotation pins. Click a pin to comment; hover a card to highlight it.
Step-through interactive HTML prototypes with per-step commenting, zoom, and pan support.
A single endpoint returns only open comments as structured JSON — purpose-built for LLM agent consumption.
Six reference types — section, quote, line, annotation, step, general — so every comment knows where it points.
Single static Rust binary. No runtime, no containers, no config files needed. Download and run.
A continuous cycle: humans flag issues, agents fix them, humans verify.
Place markdown, images, or HTML prototypes into the workspace folder.
Reviewers browse in the split-pane UI and leave anchored comments.
Agents poll /api/export/{id} for open comments.
Agents process feedback, make changes, and mark comments resolved.
The export API returns only what agents need — open comments with anchors and metadata.
Every comment carries a typed reference so agents know exactly where to look and what to fix.
section — markdown heading anchorquote — inline text selectionline — line numberannotation — image pin numberstep — prototype step indexgeneral — whole-item commentComments are never lost in context — every anchor type maps precisely to a location.
Heading anchor
Text selection
Line number
Image pin
Prototype step
Whole item
A production-ready Rust binary for the review API, a Vue frontend for the review UI, and a vendored `common/` base for shared docs/demo/release entry points.
Start reviewing with humans and agents today. No accounts, no cloud — just drop files and go.