Claude Code team shares how it runs long-lived agent loops with /loop and /goal
petergyang · x · 2026-07-21
Key takeaways from Claude Code’s loop workflow
Thariq from the Claude Code team explains how they use /loop, /goal, and workflows to run long-lived agent sessions differently.
- Before building, ask Claude to look for blind spots in the plan and surface unknown unknowns in the codebase.
- The team says they cut Claude Code’s system prompt by 80% as models got smarter and needed less direction.
- Planning is described as an iterative process of exploring, investigating, and discovering what you don’t know.
- The episode also covers a live demo of editing launch videos with Claude and using HTML artifacts to plan and learn.
Related event: Claude Code Team Shares Long-Running Agent Workflows(5 posts)→
More from coding & agent
- Hugging Face ships hf CLI updates: deduped cache saves 19GB, faster buckets, retryable GPU jobs — yb2698 · 2026-10-26
- Fern 0.1 debuts: a model built to audit training data and debug agent traces for AI builders — bneyshabur · 2026-10-26
- Fern 0.1 posts 73.4 on SWE-Bench Verified, far behind frontier models on agentic benchmarks — bneyshabur · 2026-10-26
- One prompt, full Blender animation: dev says he never touched Blender once — iamfakhrealam · 2026-10-26
- Claude Code finds root backdoor in Honda Civic head unit, then debloats it — IndraVahan · 2026-10-26
- Claude Code + Monid MCP builds a $12K/year B2B prospecting engine on live data — aryanXmahajan · 2026-10-26