Claude Cowork + LMS Integration: Connecting to Cornerstone, Docebo and Workday Learning

Why Claude Cowork + LMS Integration Matters

Learning and development teams face a critical bottleneck: content generation takes weeks, but deployment windows are hours. When your LMS sits disconnected from your content creation pipeline, L&D managers manually copy, paste, and reformat course materials across systems. This creates versioning chaos, broken links, and courses that ship with outdated information.

Claude Cowork changes this equation. By integrating Claude Cowork directly with enterprise LMS platforms like Cornerstone OnDemand, Docebo, and Workday Learning, you can:

• Generate course content, assessments, and SCORM modules in hours instead of weeks
• Publish directly to your LMS without manual re-formatting or file conversions
• Maintain compliance, SSO governance, and role-based access controls across the pipeline
• Scale content authoring across multiple teams while enforcing organizational standards
• Reduce content creation costs by 70% through AI assisted authoring

Companies with mature Claude Cowork implementations report cutting course development cycles from 8 weeks to 2 weeks for new training programs. What once required 3 instructional designers now requires 1 plus strategic oversight. That frees L&D teams to focus on curriculum strategy instead of content assembly.

Connecting Claude Cowork to Cornerstone OnDemand

Cornerstone OnDemand is the market leader in enterprise LMS deployments. Most Fortune 500 companies use Cornerstone for compliance training, onboarding, and skills development. Integrating Claude Cowork with Cornerstone means connecting through a combination of REST APIs and MCP (Model Context Protocol) servers.

Cornerstone API Architecture

Cornerstone exposes its content authoring pipeline through OAuth 2.0 REST APIs. The integration flow looks like this:

1. Claude Cowork calls your custom MCP server with course specifications
2. Your MCP server generates course structure, modules, and content via Claude
3. The MCP server converts generated content to SCORM 2004 or xAPI format
4. Your backend calls Cornerstone's Content Upload API with authentication
5. Cornerstone ingests the package and creates a draft course
6. L&D managers review and publish from Cornerstone's native interface

This approach gives L&D teams AI-assisted content generation while maintaining full governance and compliance controls within Cornerstone. No generated course goes live without human approval.

Setting Up Cornerstone OAuth Credentials

Your Cornerstone administrator needs to create OAuth credentials in the Cornerstone admin portal. Request:

• Client ID and Client Secret for your integration account
• OAuth endpoint URL (typically https://csod.com/oauth/authorize)
• API base URL for your Cornerstone tenant
• Content creator role with course upload and SCORM package permissions

Store these credentials in your application's secure secrets manager (AWS Secrets Manager, Azure Key Vault, or HashiCorp Vault). Never hardcode credentials in your MCP server code.

Building Your Cornerstone MCP Server

Your MCP server acts as the bridge between Claude Cowork and Cornerstone. It should expose two main functions:

• GenerateCourseContent: Takes course title, duration, learning objectives, and target audience. Returns structured JSON with module outlines, lesson content, and assessment questions.
• PublishToCornerstone: Takes the generated content, converts to SCORM, and uploads to your Cornerstone tenant via the Content Upload API.

The MCP server should also validate that all uploaded content meets your organization's standards: proper learning objective alignment, assessment validation, accessibility compliance, and brand consistency.

Docebo REST API Integration for Content Publishing

Docebo is the fastest-growing cloud LMS, popular with mid-market and enterprise organizations scaling global training. Docebo's REST APIs are more modern and developer-friendly than Cornerstone, making it an excellent platform for Claude Cowork integration.

Docebo API Key Setup

In your Docebo admin panel, navigate to Integrations > API and generate a new API key. Docebo uses token-based authentication rather than OAuth. Your API key should have these scopes:

• courses:create and courses:update for creating and editing courses
• courses:publish for course publication workflows
• courses:upload for SCORM and xAPI package uploads
• users:view for role and permission checking
• reports:view for enrollment and completion tracking

Store your Docebo API key as an environment variable and never include it in version control.

Content Publishing Workflow with Docebo

The typical Docebo publishing workflow with Claude Cowork:

1. Claude Cowork generates course content and lesson structure
2. Your MCP server calls Docebo's Create Course endpoint with metadata (title, description, category)
3. Docebo returns a course ID
4. Your MCP server generates SCORM 2004 package from Claude's output
5. Upload the SCORM package to the course using Docebo's Upload LO endpoint
6. Set course visibility, enrollment rules, and completion criteria
7. Docebo automatically indexes the course for search and recommendations

Docebo handles the LMS plumbing: learner enrollment, progress tracking, transcript management, and certification workflows. Your job is ensuring Claude generates course content that meets Docebo's content packaging standards.

Docebo REST API Endpoints You'll Use

Core endpoints for content authoring integration:

POST /api/v1/courses
    Request body: {
      "title": "Advanced Claude Usage",
      "description": "Learn enterprise patterns",
      "category_id": 123,
      "language": "en"
    }

POST /api/v1/courses/{courseId}/learning-objects
    Request body: {
      "title": "Module 1: Foundations",
      "type": "scorm",
      "file": <binary SCORM zip>
    }

POST /api/v1/courses/{courseId}/publish
    Request body: {
      "publish_date": "2026-04-15",
      "availability": "immediate"
    }

Workday Learning Connector Setup

Workday Learning is part of the broader Workday suite, integrated with HR, finance, and talent management. If your organization uses Workday as your HRIS, integrating Claude Cowork with Workday Learning ensures course enrollment workflows are automatically aligned with your org structure.

Workday REST Web Services Configuration

Workday uses REST Web Services for integration. You'll need:

• A Workday integration user account with Learning Content Creator role
• Web Service access configured in Workday admin (Security > API Clients)
• OAuth 2.0 client credentials (Client ID and Client Secret)
• Your Workday tenant URL and web services endpoint

Unlike Cornerstone and Docebo, Workday Learning emphasizes structural content that maps to organizational hierarchies. Courses in Workday are often tied to specific roles, departments, or competency frameworks.

Creating Learning Content in Workday

Workday Learning content is stored as Learning Content objects. When Claude Cowork generates content for Workday, structure it with these elements:

• Learning Content ID (unique identifier, often a slug like "2026-Q1-Claude-Certification")
• Learning Content Type (e.g., "Online Course", "Virtual Instructor-Led", "Self-Study")
• Content metadata: title, description, learning objectives, duration
• Competencies and skills the course teaches
• Assessment strategy (pre-assessment, post-assessment, or both)

Your MCP server should generate these structured data elements from Claude's output, then POST them to Workday's Learning Content Creation REST endpoint.

Workday Enrollment Integration

The power of Workday Learning integration is automatic enrollment. Once your course is published, you can create Learning Assignments that automatically enroll users based on:

• Organizational department or cost center
• Job role or job family
• Competency gap (users who don't yet have a required skill)
• Mandatory compliance training (triggered by hire date or department)

This means Claude Cowork can generate a compliance training course, your MCP server publishes it to Workday, and Workday automatically enrolls 10,000 employees based on their role. No manual enrollment lists required.

SCORM Content Publishing Workflows

SCORM (Sharable Content Object Reference Model) is the standard format for LMS-agnostic course packaging. All three platforms (Cornerstone, Docebo, Workday) accept SCORM 2004 packages. Understanding how to convert Claude-generated content to valid SCORM is critical for enterprise integration.

SCORM Package Structure

A valid SCORM 2004 package is a ZIP file with this structure:

course-package.zip
  ├─ imsmanifest.xml (metadata and structure)
  ├─ content/
  │  ├─ lesson-1.html
  │  ├─ lesson-2.html
  │  ├─ assets/
  │  │  ├─ course-banner.png
  │  │  ├─ infographic.svg
  │  │  └─ intro-video.mp4
  │  └─ SCORMAdapter.js (communication layer)
  └─ metadata/
     └─ LMS-specific tracking data

The imsmanifest.xml file is the key. It contains:

• Course metadata (title, description, learning objectives)
• Module and lesson structure (hierarchical organization)
• SCO (Sharable Content Object) references
• Mastery score and completion criteria
• Launch parameters and runtime configuration

Converting Claude Output to SCORM

Your MCP server should implement a SCORM converter that takes Claude's JSON course outline and generates valid imsmanifest.xml plus HTML lesson files. A typical conversion pipeline:

1. Claude generates JSON with course structure: modules, lessons, objectives, assessments
2. Your MCP server parses the JSON and extracts learning objectives and mastery criteria
3. Convert lesson content to semantic HTML5 with accessibility attributes
4. Generate SCORMAdapter.js with proper cmi.core.score tracking
5. Build the imsmanifest.xml with proper namespace declarations
6. Package everything as a ZIP file with SCORM compliance validation

After generating the SCORM package, validate it using a SCORM validator (e.g., ADL SCORM Conformance Test Suite) before uploading to your LMS. Invalid SCORM packages cause courses to fail on import.

SCORM Tracking and Completion

SCORM tracks learner progress through JavaScript communication between the course HTML and the LMS. When a learner completes a quiz, submits an assessment, or reaches a section of content, the course calls:

// Set learner's score
scormAdapter.setValue("cmi.core.score.raw", 85);

// Mark lesson complete
scormAdapter.setValue("cmi.core.lesson_status", "completed");

// Record time spent
scormAdapter.setValue("cmi.core.session_time", "PT45M00S");

// Commit data to LMS
scormAdapter.commit();

Your Claude-generated content should include these SCORM calls at appropriate moments: after each assessment, when learners complete a module, and at course exit.

Admin Governance and SSO Considerations

Enterprise LMS integrations require governance layers that prevent unauthorized content publishing and ensure compliance. Claude Cowork with LMS integration must include:

Role-Based Access Controls

Define clear roles in your integration:

• L&D Content Creator: Can request Claude to generate course content but cannot publish without approval
• L&D Manager: Can review and approve generated content for publishing
• Compliance Officer: Can mark courses as compliant or flag for revision
• LMS Administrator: Can configure integration parameters and audit publishing activity

Your MCP server should enforce these roles before content is published to your LMS. A content creator cannot directly call the Docebo or Cornerstone publish endpoint.

SSO Integration

If your organization uses enterprise SSO (Okta, Azure AD, Ping Identity), integrate your MCP server's authentication with your identity provider. This ensures:

• Users authenticate once and automatically have access to Claude Cowork and the publishing interface
• Roles and permissions flow from your identity provider (no separate user management)
• Audit logs capture which user published which course and when
• Revoked users immediately lose access to content authoring tools

Most LMS platforms support SAML 2.0 or OpenID Connect. Configure your MCP server to verify JWT tokens from your identity provider and include user claims (name, email, roles) in all LMS API calls.

Content Audit and Compliance

Every course published via Claude Cowork should be auditable. Log:

• The original prompt or specifications Claude received
• The exact Claude model version used (Sonnet 4.6, Opus 4.6, etc.)
• Who requested the content, who approved it, and when
• The generated content (imsmanifest.xml, lesson HTML, assessments)
• Which LMS the content was published to and the upload timestamp

Store these audit logs in an immutable log system (AWS CloudTrail, Splunk, or ELK) so compliance auditors can trace content lineage.

The LMS Integration Checklist

Use this workflow to implement Claude Cowork + LMS integration in your organization:

Prompt Templates for LMS Content Generation

Use these Claude prompts to generate LMS-ready content that integrates seamlessly with your publishing workflow.

Prompt Template 1: Generate Course Structure

You are an instructional designer creating course content for [LMS_NAME].

Create a complete course structure for:
Course Title: [TITLE]
Target Audience: [AUDIENCE]
Duration: [MINUTES]
Learning Objectives: [OBJECTIVES]
Required Knowledge: [PREREQUISITES]

Output a JSON object with this structure:
{
  "course_title": "...",
  "learning_objectives": ["...", "...", "..."],
  "modules": [
    {
      "module_title": "...",
      "duration": 60,
      "lessons": [
        {
          "lesson_title": "...",
          "content": "detailed markdown content",
          "key_takeaways": ["...", "...", "..."],
          "assessment": {
            "type": "multiple_choice",
            "questions": [
              {
                "question": "...",
                "options": ["...", "...", "...", "..."],
                "correct_answer": 0,
                "explanation": "..."
              }
            ]
          }
        }
      ]
    }
  ],
  "mastery_score": 80,
  "completion_criteria": "All lessons completed + 80% on final assessment"
}

Follow these rules:
  Ensure all lessons include specific, measurable learning outcomes
  Include 2-3 formative assessments per module
  Write assessments to test retention and application, not recall
  Keep lesson content between 300-500 words
  Include real-world examples relevant to [AUDIENCE]

Prompt Template 2: Generate SCORM-Compatible Lesson HTML

Convert this lesson content to SCORM 2004-compliant HTML:

Lesson Title: [LESSON_TITLE]
Content: [CONTENT_MARKDOWN]
Learning Objectives: [OBJECTIVES]
Assessment Questions: [QUESTIONS_JSON]

Generate semantic HTML5 with:
  - ARIA labels for accessibility
  - Progress tracking hooks (SCORM cmi.core.lesson_status)
  - Assessment submission handlers
  - Mobile-responsive design (works on phones, tablets, desktops)
  - Inline styling (no external CSS files)

Include a footer with SCORM initialization:
<script>
  const scormAdapter = initializeSCORM();
  document.addEventListener('DOMContentLoaded', () => {
    // Track lesson viewed
    scormAdapter.setValue('cmi.core.lesson_status', 'incomplete');
    scormAdapter.commit();

    // Track assessment completion
    document.getElementById('submitButton').addEventListener('click', () => {
      const score = calculateScore();
      scormAdapter.setValue('cmi.core.score.raw', score);
      if (score >= 80) {
        scormAdapter.setValue('cmi.core.lesson_status', 'completed');
      }
      scormAdapter.commit();
    });
  });
</script>

Prompt Template 3: Generate Assessment Content

Create a final assessment for this course:

Course: [COURSE_TITLE]
Learning Objectives: [OBJECTIVES]
Difficulty Level: [BEGINNER/INTERMEDIATE/ADVANCED]

Generate a JSON object with 10-15 questions covering:
  - 40% recall of key concepts
  - 40% application to real-world scenarios
  - 20% synthesis and critical thinking

Format:
{
  "assessment_name": "...",
  "passing_score": 80,
  "time_limit_minutes": 30,
  "questions": [
    {
      "id": "q1",
      "type": "multiple_choice",
      "question": "...",
      "options": ["...", "...", "...", "..."],
      "correct_answer": "...",
      "explanation": "why this is correct",
      "cognitive_level": "remember|understand|apply|analyze"
    }
  ]
}

Ensure questions test understanding, not memorization. Include scenario-based questions.

Key Takeaways

Related Articles