For teams working with AI agents

Turn team feedback into better AI-generated work.

Bring your team into the review, agree on what matters, and give your AI agent a clear next step.

Reviewable is one shared place to look at what your AI generated, turn everyone's comments into clear decisions, and move the next version forward.

Reviewing a sample opens a temporary review you own: comment, decide, and build the handoff yourself. It expires after seven days and cannot be recovered. The finished review is a shared example, reseeded from time to time — you can comment on it, but only its owner can decide. Read the protocol.

A Reviewable review moving from an artifact through an anchored comment and approved decisions into the R handoff mark.

How a review works

Your team reviews and decides in Reviewable. Your agent generates and regenerates outside it, working from what you approved.

Your agent, outside Reviewable

01

Generate

Your agent produces the work in its own environment.

In Reviewable — where your team takes control

02

Review

Your team looks at the work and comments directly on what they see.

03

Decide

You accept or discard each proposed change, one by one.

04

Handoff

What you approved becomes a clean, fixed summary — nothing else.

Your agent, with your decisions

05

Regenerate

Your agent reads that summary and produces the next version.

Why the feedback actually lands

Reviewable does one thing: give your team real control over what gets decided, then make that decision usable by your agent.

Anchored decisions

Comments attach to the exact part of the work they are about, and each one is accepted or discarded on purpose.

A clean handoff

Your agent receives what was agreed — not a transcript to dig through.

Works with your agent

Your agent reads the reviews you share with it, and can publish a new one only after you approve a scoped connection.

No accounts, no lock-in

Share by link, try the public demo, and read the whole protocol for yourself.

A Reviewable review: two comments pinned to a campaign draft, one accepted and one still needing a decision, beside the handoff those decisions produce.

Hand off decisions, not chat history

When you approve changes, Reviewable turns those decisions into clear, stable context your agent can use for its next pass.

Read the agent-context/v1 specification
An approved handoff card carrying owner decisions into the Reviewable R mark.

Add Reviewable to your agent

Give your agent a direct way to find live reviews, read approved handoffs, and publish a new artifact for review without copying context by hand.

No workspace connection is needed yet. Install Reviewable now; when your agent needs review data, it starts a browser authorization. You approve a scoped connection without giving the agent your workspace key.

The guided path installs the MCP and the make-it-reviewable action together, for Codex, Claude Code, or any stdio MCP host.

After installation, ask your agent

  • “What reviews are ready for me to work on?”
  • “Read the latest Reviewable handoff before you make changes.”

Or add the MCP directly

Codex

codex mcp add reviewable -- npx -y reviewable-artifacts-mcp

Claude Code

claude mcp add --transport stdio reviewable -- npx -y reviewable-artifacts-mcp

Using another agent? Read the agent onboarding guide. Reviewable is also available from npm and the MCP Registry.