MCP server

MCP (Model Context Protocol) is the open standard that lets AI agents and editors call external tools. Connect this server and your agent can search looprepo's vetted loop catalog, grade any loop prompt A–F, and generate safe-loop harnesses — no auth, no API key, rate limit 60 req/min per IP.

One-click install

Endpoint: https://mcp.looprepo.dev/mcp

Add to Claude →Add to Cursor →Add to VS Code →VS Code Insiders →

The Claude button opens Claude Desktop directly if it’s installed, otherwise claude.ai in your browser — either way, go to Customize → Connectors and click “+ Add custom connector” to paste the endpoint above (Claude has no one-click MCP install). Cursor opens the app if it’s installed, otherwise Cursor's install page; the VS Code buttons go through vscode.dev’s install redirect. Everything else below is copy-paste.

Claude Code

claude mcp add --transport http looprepo https://mcp.looprepo.dev/mcp

Claude Desktop

Customize → Connectors → “+ Add custom connector” with the endpoint URL above — or add this to claude_desktop_config.json (uses the mcp-remote bridge; needs Node installed):

{
  "mcpServers": {
    "looprepo": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://mcp.looprepo.dev/mcp"]
    }
  }
}

Codex CLI

codex mcp add looprepo --url https://mcp.looprepo.dev/mcp

Or add it to ~/.codex/config.toml directly — the same config drives the Codex IDE extension:

[mcp_servers.looprepo]
url = "https://mcp.looprepo.dev/mcp"

Windsurf

Add to ~/.codeium/windsurf/mcp_config.json:

{
  "mcpServers": {
    "looprepo": { "serverUrl": "https://mcp.looprepo.dev/mcp" }
  }
}

Any other MCP client

Generic Streamable HTTP config (Cline, Zed, custom SDK clients, …):

{
  "mcpServers": {
    "looprepo": { "type": "http", "url": "https://mcp.looprepo.dev/mcp" }
  }
}

Tool catalog

22 tools across four groups. Read tools are free; the two LLM-backed cost centers — crafting loops and building harnesses — are daily-capped (see limits below).

Tool	What it does	Cost
Discovery
loops_search	search the catalog by text, category, loop type, safety grade, or provenance	free
loops_get	fetch one loop’s full prompt, notes, attribution, and stats by slug	free
loops_categories	list categories and loop types with live published counts	free
loops_trendingnew	surface loops gaining copies and upvotes right now	free
loops_similarnew	find loops close to a given slug by category, type, and tags	free
loops_for_stacknew	match loops to a described stack — framework, tools, agents	free
loops_comparenew	put two or more loops side by side on safety, scope, and cost	free
loops_statsnew	directory totals — counts by grade, category, and loop type	free
loops_of_the_weeknew	the current Loop of the Week — top-voted, refreshed weekly	free
Grading & safety
loops_grade	grade any loop prompt A–F with the deterministic, default-fail Safety Score	free
loops_hardennew	rewrite a loop to add missing exit conditions, budgets, and gates	LLM · capped
loops_grade_explainnew	plain-language breakdown of why a loop earned its grade	LLM · capped
harness_validatenew	check a harness bundle for the required safety scaffolding	free
Build & intake
loops_generate_harness	generate a complete safe-loop harness bundle for a goal	LLM · capped
loops_intake_recommend	turn a read-only Claude-use report into ranked real loop matches plus a builder prefill	free
loops_craftnew	draft a net-new candidate loop from a described workflow, staged through the same gate	LLM · capped
intake_prompt_getnew	return the read-only Claude-use intake prompt to paste into an agent	free
loops_submitnew	submit a loop to the review queue — text only, sanitized, gated	free
Account
cme_tier_statusnew	show separate Looprepo rolling allowances and CME plan entitlements; the optional verified account email adds its build counter	free
beacon_run_reportnew	report a loop run to Mission Control telemetry	free
articles_searchnew	search looprepo’s published loop-engineering articles	free
articles_getnew	fetch one article’s full text by slug	free

Not sure what to automate? Start with Claude intake to collect a read-only evidence report before choosing a loop.

Access & rate limits

No auth, no API key, no signup — all read tools are open.
60 requests/min per IP across the API (429 + Retry-After when exceeded).
Request bodies capped at 32KB (413 above that).
generate_harness additionally: 30 builds/day per email, per-IP cap.

Daily limits

Only the two LLM cost centers are metered — crafting loops and building harnesses. Read tools have no per-day cap.

Call cme_tier_status with an optional CME API key and matching verified CME account email to see remaining network, hourly, and email allowances separately from CME account-plan entitlements.

Tier	Crafted loops / day	Harness builds / day
Anonymous	20	10
CME Pro	100	30
CME Power	300	100

Higher limits for ConnectMyEmail members →

Troubleshooting

405 Method Not Allowed. You hit an old deploy or sent a GET. The transport is POST-only JSON-RPC at mcp.looprepo.dev/mcp. If a client saw 405s before 2026-07-06, retry — the fix is deployed.

403 from Cloudflare. Cloudflare Bot Fight Mode used to block Python-based MCP clients by user-agent. That is resolved. If you still see a 403 with a cf-ray header, tell us — it never reached the server.

“It returns HTML”. looprepo.dev/mcp (this page) is documentation. The MCP endpoint your client connects to is https://mcp.looprepo.dev/mcp — a JSON-RPC transport, not a web page.

Endpoints

https://mcp.looprepo.dev/mcp

MCP transport (Streamable HTTP, JSON-RPC 2.0) — POST only

https://mcp.looprepo.dev/health

health check

Prefer plain HTTP? The catalog is also directly fetchable — see the agents page for catalog.json, catalog.txt and llms.txt. Want to check a loop before running it? Grade it →