SKILLJAR: MCP use cases?!

​@its_katkenny I was not at Pulse, but have been excitedly awaiting this functionality for Skilljar. It is very intriguing and we are looking at a few initial use cases:

• Deeper MoM reporting and learner analysis
• Updating content via Claude - titles, descriptions, etc.
• Assistance with localization tasks for course families (pulling the information in English and then putting it back in translated language)
• Bulk tasks - creation of a number of courses for an acquisition project, cleaning up old lessons or archived assets

Unfortunately, the requirement at the moment is Claude Code. While our organization has that, it’s reserved for developers. So we will anxiously wait for Claude Desktop to be supported (it’s coming from all that I see in the support information). 

​@its_katkenny  I’m using Claude Code to vibe-code a Skilljar content editing tool for our team. It currently uses the Skilljar v1 API, but I plan to switch it over to MCP soon. It can read and write Skilljar HTML lessons. Features I am working on:

• Training content update suggestions based on software changes from Jira
  • Uses a Google Docs/track changes-style UI for human accept/reject
• An “apply skill” workflow to bulk-apply Claude skills to our content, also using the suggestion UI
  • Example skill: rewrite this exercise using our new exercise template
• Global find-and-replace
• Bulk quiz and question editing
• A WYSIWYG editor so we don't have to go back to Skilljar, including fixing several things we find annoying about the existing Skilljar UX
• A GitHub-based content backup system

​@sam.walker  LOVE it ! are you using SCORM?  I’ve been some pretty awesome courses in Claude Design that I export and import SCORM into Skilljar but that’s very time consuming each time i want to make a quick update to a course.  Or how have you done this?

Hey ​@its_katkenny, thanks! It’s been a fun project. I’m excited to see where we can take it.

No, we do not use SCORM. I tried to clarify that above, “Skilljar HTML lessons.” That is where all our text and image content lives, which is the majority of our courses.

But if SCORM uploads are a pain point for you, I imagine you could use the Skilljar API or MCP to sync them more easily. I haven’t used the API for uploading SCORM, but I assume that is what Partio.io does (I’ve seen their “sync content to Skilljar” workflow demo). We use the v1 API to upload HTML lessons. I couldn’t figure out a good workflow for image uploads using the Skilljar API, so we just use the AWS CLI for that.

I always advise explaining the use case to your agent first, rather than assuming a good solution off the bat. But it sounds like you could probably have the agent listen for commit webhooks if you use GitHub, or have it write a script for you that lives in your content root and pushes changed content automatically, or on a user trigger.

The flow I am building right now (partly working) looks like this:

• Our AI Assistant tool lives on a server
• Users (the Customer Education team) log in to it
• Users can browse and open any Skilljar HTML lessons, organized in learning paths and courses
• Users can apply skills to one or more lessons
  • Skills can run as Suggestions (human-approved) or Auto-Update (changes applied automatically)
• Once a skill runs, it generates Track Changes-style suggestions with “Accept” and “Reject” buttons
  • The user can go through the content and manually accept or reject these changes, then save the content to confirm the changes
• Once the user has updated the content, they can select lessons to “push” to Skilljar. The tool checks the versions of everything and confirms what will be overwritten and what will be new. If it all looks correct, the user can push it.
  • We have a different tool that automatically backs up our Skilljar content to GitHub in case things go sideways here.