Adding claude-pipe to Development & Code Tools.

When building multi-step pipelines with Claude Code — research, batch processing, overnight generation jobs — you need state persistence, failure recovery, and worker coordination that Claude Code doesn't give you out of the box.

claude-pipe fills those gaps with conventions: file-based YAML state you can cat, flat orchestration where ROOT controls everything, data layer contracts between phases, and resume that picks up from the exact failure point.

Two runnable examples: research pipeline (plan then fan-out research then synthesize then report) and batch classifier (partition then classify in parallel then aggregate).