The Agentic Developer: Orchestrating AI Workflows with MCPs
summary
Claim: one prompt triggers an add content skill that uses MCP servers for Cloudinary and GitHub, Nuxt content markdown generation, and a Playwright-based verification loop that runs end to end and creates a PR. Tools actually used include Copilot/Cloud agents, the Playwright CLI skill (token‑efficient), Remotion for video rendering and Eleven Labs MCP for a voice clone. A release‑video pipeline takes about 30 minutes to produce after ~6 hours of skill authoring, and it works in real runs.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 3:12Manual website updates problem
- 5:01AI content workflow demo
- 8:55Agentic workflow operation
- 10:19AI skills importance
- 13:25Skill structure breakdown
- 15:55Global vs project skills
- 16:57Finding community skills
- 18:22MCP servers vs skills
- 19:24Key takeaways
- 20:07Automating release videos
- 24:40AI-generated release video demo
- 27:08Summary and call to action
related talks
16mDriving agentic innovation w/ MCP as the backbone of tool-aware AI
12mMCP and the Dawn of Agentic Intelligence
20mAI Agent Skills and MCP: Better Together
48mSupercharged Testing: AI-Powered Workflows with Playwright + MCP
21mBuilding Agents (the hard parts!)
01h 28mAI Coding Agents and how to code them
37mLetting AI Interface with Your App with MCPs
32mFigma deep dive: Agentic workflows and the MCP