Anthropic Workshop: Building a Self-Grading Research Agent on Claude Managed Agents
AI Engineer · youtube · 2026-10-12
Part 2 of Anthropic's Claude Managed Agents workshop features a hands-on build led by Priyanka Phatak and Gabe Cemaj.
- Starting from a one-line console agent, they build a 'research desk' app with a chat UI on the session's event stream.
- They add an SEC EDGAR skill, subagents, a custom tool, and a shared memory store, treating skills as live, updatable artifacts.
- The core mechanism is outcomes-based self-supervision: a task plus a rubric and grader that checks Claude's Nvidia analysis and forces retries until it passes.
- Q&A covers vaults for placeholder secrets, Agent SDK vs managed agents, guardrails for agent loops, memory sharing, and MCP tunnels.
More from coding & agent
- Screego: open-source self-hosted screen sharing for devs with 10.5k GitHub stars — thisguyknowsai · 2026-10-12
- Grok app adds Bots: users can now build custom bots like Trader inside Grok — nima_owji · 2026-10-12
- Moli: open-source headless browser for AI agents renders pixels only on demand — abhishek__AI · 2026-10-12
- Open-source real-time object detection with YOLOv8, OpenCV, and threaded frame buffering — mariamamhan · 2026-10-12
- Claude Opus 5.5 makes a Blender short film about the last black hole — 3,000 lines of Python, no gen models — toni_btrain · 2026-10-12
- HF Engineer Runs pi-web on DGX Spark to Steer Local AI Agents From His Phone — lhoestq · 2026-10-12