Your next workflow.
Inside your agent.
Bring the catalog into the conversation. Find a loop, inspect its building blocks and check your changes with the tools you already use.
Codex
Claude Code
Cursor
Antigravity
OrcaMake the connection.
This static catalog does not host an MCP server. Start with the agent skill; hosted MCP connection will be available after provisioning.
Start with a real task.
Let your agent do the searching. Keep the decisions with you.
Find a starting point
Find a loop for preparing a daily research digest. Compare its inputs, components and review requirements before recommending it.
Understand a workflow
Inspect the exact version of this loop. Explain its steps, destinations and approval gates. Tell me what I need to configure.
Adapt with confidence
Read the qloops schema and adapt this loop to my task. Reuse existing components. Validate the result and explain any errors.
From intent to a reusable loop.
01 / Find
Search the catalog and choose an existing loop before creating new steps.
02 / Adapt
Your agent reads the versioned definition and makes changes in your project.
03 / Check
Validate against the pinned Core contract. Execution happens locally, after your review.
A few things to know.
Does MCP run my workflow?
No. This endpoint exposes catalog, search, get, schema and validate. It does not execute workflow steps or call a model.
Is this the same as Get started?
Get started installs qloops and gives your agent the skill. MCP adds a direct way for a compatible client to inspect and validate catalog materials.
Can a cloud agent connect to localhost?
No. This preview address is reachable only from your computer. A publicly hosted endpoint is not provisioned yet.
Is a successful check proof of a run?
No. Validation checks the definition. A real run needs its own outputs and evidence.
Which clients are verified?
The server is tested with the MCP SDK. Logos show setup targets, not a claim that every client has completed an end-to-end connection.