Podcast/Agentic AI/ Is Your Team’s Claude Code Setup the Wild West? Plugins Fix That

Agentic AI

Episode 68 · Season 1

Is Your Team’s Claude Code Setup the Wild West? Plugins Fix That

August 14, 202630:09

The moment one person on your team builds a genuinely useful agent setup in Claude Code or Claude Cowork, you hit a new problem: sharing it without it splintering into a dozen diverging copies. Cirrius Solutions COO Tim Harting and Head of Custom Development Gavin Franklin break down plugins – the packaged, version-controlled way to ship one standard agent configuration to a whole team and govern it as it grows.

What you’ll learn:

  • What goes into a plugin – skills, commands, agents, hooks, and MCP services under a plugin.json manifest – and how a Git repo turns it into an easy, auto-refreshing team install
  • A three-tier governance structure: a company-wide core plugin, department-level plugins, and an individual layer whose proven ideas feed back into the next release
  • How Claude plugins compare with OpenAI’s Codex approach, and why roughly 90% of a Claude plugin runs there as-is

Who it’s for:

Technology and business leaders rolling agentic tools out past the first power user – CTOs, engineering and platform leads, and department heads who want standardization without shutting down individual creativity.

Guest + credibility:

Tim Harting, Cirrius Solutions’ COO, has led the company’s internal plugin rollout – a company-wide ops agent plus department plugins for sales, managed services, project delivery, and custom development – and calls plugins one of the highest value, lowest lift things that you could probably start tomorrow. Gavin Franklin, Cirrius Solutions’ Head of Custom Development, walks through the build side: converting a working local configuration into a plugin, testing it, and publishing versions through Git.

Primary CTA:

Follow Cirrius Talk: Agentic AI, then download the episode artifact linked below to see the full plugin structure and three-tier governance model on one page.

Recommended Links & Resources:

Any Questions: Reach out to us at CirriusTalk@cirriussolutions.com

We cover

  • Understanding Plugins and Their Importance
  • Creating and Managing Plugins
  • Three-Tier Governance Structure for Plugins
  • Project vs User Level Plugin Installation
  • Differences Between OpenAI and Anthropic Plugins
  • Best Practices for Plugin Distribution

Chapters

  • 0:00Introduction to Agentic AI and Plugins
  • 0:38Introduction to AI Governance and Plugins
  • 3:00Understanding Plugins and Their Importance
  • 5:43Creating and Managing Plugins
  • 8:29Three-Tier Governance Structure for Plugins
  • 11:04Project vs User Level Plugin Installation
  • 13:49Differences Between OpenAI and Anthropic Plugins
  • 16:32Best Practices for Plugin Distribution
  • 18:51The Role of Governance in AI Tools
  • 21:33Real-World Applications and Use Cases
  • 24:28Conclusion and Future Considerations
  • 29:01Introduction to Governance Structure
  • 29:28The Challenge of Implementation
  • 29:54Closing Remarks and Future Engagement

Resources and downloads

The artifacts Greg’s team built for this episode, free to download.

Claude Code Plugins: Standardizing Agentic Rollouts Across Your Team

Walks the full plugin lifecycle from scaffold to team-wide auto-update: what actually lives inside a Claude Code plugin (skills, commands, agents, hooks, and MCP connections), why one versioned git artifact beats copy-pasted prompts and snowflake configs, and how a private marketplace repo powers three-tier governance where proven skills graduate upward by pull request. Closes with a side-by-side of the OpenAI Codex equivalents and where the two plugin models converge.