This revised prototype is no longer just a directory. It now tries to answer three first-order questions for a newcomer: what this ecosystem is, which projects shape it, and how to get started without over-installing. It still includes the full 226-item seed index plus 10 deeply enriched exemplars.
The ecosystem is not one monolithic tool landscape. It behaves more like a stack: foundational docs and defaults, workflow/methodology layers, orchestration systems, observability surfaces, and safety/quality guardrails.
The list categories are useful for browsing, but the more helpful model is a stack: official defaults → workflow systems → orchestration → observability → guardrails.
The stable center of gravity is still Anthropic’s own guidance: start with a real repo, use clear plans, and give the agent a way to verify success.
Representative projects: Anthropic best practices, Claude Code docs mirrors
Projects like Superpowers, AB Method, and CCPM turn Claude Code into a repeatable process rather than a prompt box. This is where planning, specs, missions, and review discipline live.
Representative projects: Superpowers, AB Method, CCPM
AgentSys and Claude Squad show the “many agents, isolated workspaces” branch of the ecosystem. These tools matter once Claude Code is no longer a single-user assistant but part of a larger delivery system.
Representative projects: AgentSys, Claude Squad
As sessions get longer, users need search, replay, context awareness, and statusline feedback. Claude HUD and claude-code-tools are representative of that operational layer.
Representative projects: Claude HUD, claude-code-tools
TDD Guard and similar hooks reflect a shift toward enforced quality gates: test-first flows, explicit checks, and safer approval patterns.
Representative projects: TDD Guard, Claude Code Agents
It captures the core opinionated pattern that shows up again and again in this ecosystem: slow down, clarify, plan, test, and review instead of asking Claude Code to freestyle.
Start here if: Users who want a disciplined default way to work with Claude Code.
It shows the ecosystem’s stronger workflow-centric branch: mission-oriented work, explicit specs, and narrow execution loops.
Start here if: Builders who like specs, bounded missions, and structured delivery.
It is one of the clearest examples of people packaging Claude Code into a startup-ready set of reusable agent roles.
Start here if: Solo founders or small teams who want pre-made roles right away.
It represents the “platform layer” of the ecosystem, where Claude Code becomes one part of a larger automation and review system.
Start here if: Teams building repeatable, policy-heavy, multi-agent delivery systems.
It is a good example of bringing project management, issues, and traceability directly into agentic coding workflows.
Start here if: Teams that want backlog / issue structure tied tightly to agent work.
It addresses one of the biggest real-world pain points: resuming work, searching session history, and keeping long-running usage sane.
Start here if: Anyone whose sessions are getting longer and messier.
The dedicated JSON includes a fuller major-project list; the artifact directory still contains the detailed profiles.
This site still preserves the earlier comparison-heavy work: