A continuous accordion-fold paper sequence held by guide rails with a cobalt learning tab.
CREATE SOMETHING Learn

Build workflows operators can run and explain

Practical paths for business owners becoming operators. Start in the OpenAI ecosystem with the Codex app, create a RapidAPI-backed MCP, then use Canon image rules to make boundaries, policy, proof, and handoff visible.

Start lesson 1 Review paths
  • Learning paths 2 / 11 lessons
  • Working surface Codex app + MCP
  • Evidence Canon workflow images
  • Outcome Operator-ready proof
Course outline

Prompt. Create. Prove.

Two operator paths create the workflow and make it visible; the third scene shows the artifacts that prove another operator can run and explain it.

01 / 03 6 lessons · Learn Codex by Creating an MCP for Business Data

Build Your First Business MCP

One practical path: use the Codex app to turn a RapidAPI business-data endpoint into a stable local MCP server for a real operator workflow.

  1. 01 · Prompt Start in the Codex app

    Use the Codex app and its MCP-building skill to turn a concrete operator question into a narrow tool contract.

  2. 02 · Create Wrap the data source

    Create the MCP server, connect one RapidAPI endpoint, and keep the schema inspectable.

  3. 03 · Prove Make the work visible

    Use Canon image rules to show the object, boundary, policy gate, receipt, owner, and next action.

  1. 01
    What Codex Uses MCP For

    Understand how Codex uses MCP tools and its MCP-building skill to make business workflows repeatable.

    10 min
  2. 02
    Scaffold an MCP Server

    Use the Codex MCP-building skill to plan a minimal TypeScript server, then scaffold it locally.

    20 min
  3. 03
    Add a Business Search Tool

    Connect RapidAPI Local Business Data with Zod input validation and machine-readable output.

    25 min
  4. 04
    Connect the Server to Codex

    Register the server in the Codex app, inspect the MCP settings, and invoke the tool from chat.

    15 min
  5. 05
    Test, Debug, Iterate

    Use Inspector, Codex config checks, and stderr logs to diagnose API, schema, and prompt failures.

    25 min
  6. 06
    Ship and Extend

    Document the tool contract, safety model, evidence loop, and next useful tools.

    10 min
Receipts
tool contractlocal config
Open Build Your First Business MCP
02 / 03 5 lessons · Learn Canon by Turning Work Into Proof Images

Make Your Workflow Visible

Use Canon image rules to turn an operator workflow into clear maps, MCP boundaries, policy gates, validation receipts, and handoff artifacts.

  1. 01
    What a Workflow Image Must Prove

    Learn the Canon rule for useful images: show the object, decision state, proof, or owner.

    10 min
  2. 02
    Map Database, Automation, Judgment

    Turn a business workflow into three visible lanes so Codex and operators know what is data, execution, and policy.

    15 min
  3. 03
    Show the MCP Boundary

    Draw the edge between Codex, the MCP server, external APIs, and the human approval point.

    15 min
  4. 04
    Place Policy Gates

    Use run, review, stop, and escalate states to make automation limits visible before work ships.

    15 min
  5. 05
    Ship Receipts and Metadata

    Package the image with the claim, validation evidence, owner, refresh date, and next action.

    15 min
Receipts
workflow imagepolicy gate
Open Make Your Workflow Visible
03 / 03 4 artifacts · operator-ready

The course is judged by artifacts, not vibes.

The learning loop closes only when another operator can inspect the contract, run the server, read the workflow boundary, and name the next scoped extension.

Tool contract

Input schema, output shape, API limits, and failure behavior are explicit.

Local config

Codex can find and run the server from your environment.

Workflow image

The object, MCP boundary, policy gate, owner, and receipt are visible.

Next workflow

You leave with a scoped extension instead of a vague automation roadmap.

Evidence
  • Tool contract
  • Local config
  • Workflow image
  • Next workflow
Receipts
working MCPworkflow imagehandoff note
Review all paths
Learning handoff

Build the smallest useful workflow first.

Start with one Codex prompt, one endpoint, one schema, and one MCP call. Make the boundary visible with Canon, use .space to test it, return to .ltd for the governing principle, or carry the named workflow into .agency practice.

You are here .learn School
Owner
Learner / operator
Authority
Artifact review
Proof
Working MCP + workflow image
State
ready
  1. 01 .ltd Canon Clarify the principles, standards, and judgment that should guide the work.
  2. 02 .io Research Inspect the evidence, field reports, and operating notes behind the claim.
  3. 03 .learn School Follow a guided path and leave with an inspectable working artifact.
  4. 04 .space Workbench Run the tools and rehearse the behavior before the pattern becomes delivery.
  5. 05 .agency Build Map one workflow, qualify the control boundary, and choose the delivery path.
Open the workbenchReview the canonUse this learning in a workflow