Back to directory

Claude Proj Blueprint

A modular, level-based project template for teams building software with Claude Code, Obsidian, and Figma, letting a team pick a maturity level from a solo copilot setup up to autonomous multi-agent teams with self-healing checks.

Claude Proj Blueprint

Claude Proj Blueprint gives a team an opinionated project skeleton instead of reinventing structure from scratch, organized into four maturity levels that each include everything from the levels below: a first level of a simple copilot loop for solo projects and MVPs, a second level adding auto-invoked skills and slash commands for small teams, a third level adding automated hooks and a spec-review pipeline with multi-agent critique for teams in production, and a fourth level adding specialized agents, agent teams, and self-healing checks for mature teams.

Optional specification modules cover concerns such as compliance, security, observability, scalability, API versioning, a design system, accessibility, internationalization, testing strategy, and DevOps, and a team enables only the modules its project actually needs. The framework describes itself as a synthesis of existing industry frameworks and published research rather than an original invention, with its influences documented separately.

More in Agent skills

See all 581
  • Provenance gate

    github.com

    A deterministic authorization gateway that sits between an AI agent and its tools, tracing where each parameter of a side-effect call came from, and blocking the call or asking a human when the source does not check out.

  • A markdown framework of system rules that restates Isaac Asimov's Four Laws of Robotics as a set of prioritized loss calculations for large-language-model agents to follow, published as a discovery file for agents and crawlers to read.

  • Vitruvius

    github.com

    An engineering research agent that runs a discover, read, synthesize, verify and review loop over technical questions, reading sources directly rather than inferring from titles, and publishing its own scored verifier benchmark, weaknesses included.