Terrarium plugin for Claude Code
MCP serverAI & modelsYour AI can put the HTML pages it builds in front of you for a visual check. It publishes its work as a page you can open, you leave comments pinned to the exact spots that need changes, and it pulls that feedback in and pushes fixes. The result is pages that match what you expected before you rely on them.
Available today. Use it from your connected AI after setup.
Needs your own account with this service. Credentials stay encrypted.
After adding terrarium, ask your AI to publish an HTML page it has built. Open the page, comment where things need to change, and your AI will pull your feedback and push fixes.
Then ask your AI: use Terrarium plugin for Claude Code
What your AI can do with it
- Publish the HTML it builds as a page you can open and look at
- Pull in your review comments, each anchored to the exact spot on the page
- Push fixes in response to what you flagged
- Route its HTML work through a human visual check before calling it done
From the project's README
As published by nelsonsantryhkd/terrarium-plugin in README.md.
Give your agent a place to show its work to humans. Terrarium hosts the HTML your agent builds; humans review it in the browser with comments anchored to exact passages; the agent pulls that feedback over MCP, pushes fixes, and repeats until approved.
Install
/plugin marketplace add nelsonsantryhkd/terrarium-plugin
/plugin install terrarium@terrarium
Then mint a token at tryterrarium.com/connect
and make it available as TERRARIUM_PAT:
export TERRARIUM_PAT="samus_pat_…"
(Or just ask your agent to set Terrarium up — the bundled skill walks it through the whole flow, including sending you to /connect for the token.)
What it ships
- MCP server (
.mcp.json) — Terrarium's remote server over Streamable HTTP with 7 tools:list_artifacts,publish_artifact,get_artifact,pull_comments,push_version,reply_to_comment,resolve_comment. - Skill (
skills/terrarium/) — teaches the publish → review → fix loop, including theterrarium-feedback/1payload,expectedVersionfreshness, and comment-thread etiquette.
The protocol
pull_comments returns a versioned, structured payload
(terrarium-feedback/1) designed to be stable enough to build agents against:
review state, typed comment threads (change_request / question / bug),
text anchors with live resolution status, and reply chains. Evolution within
/1 is additive-only.
Signals
- GitHub stars
- 2
- Last commit
- Jul 2026
Advanced
- Delivery
- terrarium MCP server → your ahel gateway (mcp.ahel.ai) → every connected AI client.
- Catalog kind
- mcp-server
- Gateway key
io-github-nelsonsantryhkd-terrarium- Source
- github.com/nelsonsantryhkd/terrarium-plugin
- Hosted endpoint
https://www.tryterrarium.com/api/mcp