Claude Code adds community plugins for 32-agent orchestration and repo tours

• A community plugin called oh-my-claudecode adds a multi-agent layer on top of Claude Code, with five execution modes and claims of 3-5x faster output on multi-component builds according to the launch thread.
• The same thread says the package bundles 32 specialized agents, automatic model routing, and trigger words like "autopilot," "eco," and "plan" so teams can switch workflows without building custom orchestration from scratch the repo post.
• A separate Claude Code skill, tldr-skill, turns a repository into a self-hosted explainer site with dependency graphs, Mermaid architecture diagrams, and task-based "cookbooks," as described in the Reddit post.

The interesting shift is not a new base model. It is Claude Code getting packaged community workflows that behave more like reusable production presets. In oh-my-claudecode, those presets range from fully autonomous runs to parallel-agent execution, sequential pipelines, and a lower-cost mode. The post also describes a persistence command that keeps going until work is verified complete, plus auto-resume behavior when rate limits reset.

The installation screenshot in the thread image shows the plugin can be added through Claude Code's plugin marketplace or installed directly through npm, which makes it feel closer to a drop-in workflow layer than a custom agent stack.

tldr-skill points at a second use case: onboarding. Its author says the pipeline scans a repo locally first, then uses Claude to generate plain-English summaries, flowcharts, a searchable code map, and a cookbook for common tasks. For creative coders and small product teams, that turns repo context into something browsable and shareable instead of another private CLI session.