MCP
The Rams engine, inside your agent
One tool, review_files: your agent sends UI files and gets back a real scored design review — before the code is even committed.
Coding agents write most UI code now, and most of it ships with the same problems: clickable divs, contrast failures, magic numbers, four shades of gray that exist nowhere in your design system. The Rams MCP puts the hosted engine inside the agent writing that code. It reviews the work with all 119 versioned rules, scores it 0–100 (any critical issue caps the score at 59), and hands back concrete fixes — so the agent can address them and commit clean.
It is the same engine that reviews pull requests, so the score your agent sees locally is the score your pull request would get. The free skill is a checklist approximation; the MCP is the real thing.
Set up in two minutes
1Create an API key
In Settings → API keys, create a key. It is shown once — copy it.
2Add the server to your agent
Claude Code
claude mcp add --transport http rams https://worker.rams.ai/mcp \ --header "Authorization: Bearer rams_YOUR_KEY"
Cursor — .cursor/mcp.json
{
"mcpServers": {
"rams": {
"url": "https://worker.rams.ai/mcp",
"headers": { "Authorization": "Bearer rams_YOUR_KEY" }
}
}
}Codex — ~/.codex/config.toml
[mcp_servers.rams]
url = "https://worker.rams.ai/mcp"
http_headers = { "Authorization" = "Bearer rams_YOUR_KEY" }Any other MCP-capable client: Streamable HTTP at https://worker.rams.ai/mcp with your key as a Bearer Authorization header.
3Ask for a review
Tell your agent to run a Rams review before it commits UI changes. It calls review_files with the changed files and gets back the score and the findings:
Rams score: 59/100 (2 critical, 3 serious, 0 moderate) [critical/a11y] src/components/Banner.tsx:3 — Clickable div blocks keyboard users and screen readers entirely Fix: <a href="/promo" className="banner">…</a>
What every call returns
- A 0–100 score — any critical issue caps it at 59, so 60+ always means zero criticals
- Issues with severity, category, and file:line, from the full 119-rule engine
- A concrete fix for each finding, ready for the agent to apply
- The same judgment as your pull request reviews — one standard, everywhere
MCP reviews share your plan's monthly quota with pull request reviews. Details on scoring live in the methodology.
Common questions
How is the MCP different from the free Rams skill?
The skill is a checklist your agent reads — a free local approximation, stateless, no score. The MCP is the hosted engine itself as a tool: every call runs all 119 versioned rules with server-side verification and returns a real 0–100 score, the same review a pull request gets.
Does the MCP replace the GitHub App?
No — they cover different moments. The MCP reviews code on demand inside your agent, before it is committed. The GitHub App reviews every pull request automatically, keeps score history, and covers your whole team. Most people who use the MCP also want the App as the backstop.
How are MCP reviews billed?
Each MCP review counts toward the same monthly review quota as your pull request reviews — free 30, Solo 50, Basic 300, Pro 750 per month. There is no separate MCP charge.
What data leaves my machine?
The files you (or your agent) pass to the review_files tool are sent to the Rams worker over HTTPS, reviewed, and returned with findings. Reviews are metered against your workspace; file contents are not used to train models.
Put the engine in your agent.
Sign up free, create a key, and your agent reviews its own work — 30 reviews a month on the free plan.
Get an API keyFree