{"slug": "i-built-a-cli-that-scaffolds-agentic-workflows-for-claude-code", "title": "I built a CLI that scaffolds agentic workflows for Claude Code", "summary": "Developer Patrick Sardinha built AgentKit, a CLI tool that automatically generates orchestration files for Claude Code to manage multi-agent workflows. The tool creates markdown-based agent definitions, success criteria, and handoff procedures from a plain-language blueprint, enabling Claude Code to decompose tasks into specialized agents that run sequentially with automatic retry and escalation. AgentKit was used to build DevLog Desktop, a Tauri v2 app with local RAG via Ollama, in a single Claude Code session.", "body_md": "Most developers using Claude Code work the same way: one long conversation, manual prompting between tasks, context that grows until the model loses track of what it was doing. It works — but it doesn't scale.\n\nI built AgentKit to fix this. It's a CLI that generates the orchestration layer for any project: the files that tell Claude Code who to be, what to build, and how to hand off work between specialized agents — automatically.\n\nYou write a plain-language blueprint describing what you want to build:\n\n```\n# Todo App — Blueprint\n\n## Goal\nA simple web todo app with task creation, completion, and filtering.\n\n## Tech constraints\n- Next.js 14 + TypeScript + Tailwind\n- localStorage for persistence, no backend\n```\n\nYou run one command:\n\n```\nnpx @patricksardinha/agentkit-cli init --blueprint PROJECT_BLUEPRINT.md\n```\n\nAgentKit generates four files:\n\nThen you open Claude Code and type one instruction:\n\n```\nRead PLAYBOOK.md and execute the procedure.\n```\n\nClaude Code runs in three phases automatically.\n\n**Phase 0 — Decomposition**: it reads your blueprint and proposes a list of specialized agents. You validate before anything gets built.\n\n**Skills enrichment pause**: Claude Code creates the agent folders and waits. You can drop API docs, schemas, or any context files into the relevant agent folder before execution starts.\n\n**Phase 1 — Execution**: each agent runs in sequence, validates its success criterion, retries on failure, and escalates to you only when genuinely blocked.\n\nI used AgentKit to build DevLog Desktop — a Tauri v2 app with local RAG via Ollama.\n\nResult after one Claude Code session:\n\n**No AI inside the tool.** AgentKit is purely structural — it generates markdown files. The intelligence lives in Claude Code, not in AgentKit. This means no API key, no costs, no vendor lock-in.\n\n**Phase 0 always runs.** Most developers don't know how to optimally split work into agents. AgentKit delegates that decision to Claude Code every time, whether or not you provide a blueprint. You describe intent, Claude Code designs the structure.\n\n**Stack detection first, interactive fallback.** AgentKit reads your project directory to detect the stack. If nothing is found (new empty project), it asks you interactively from a list of supported stacks: React, Next.js, Tauri, FastAPI, Express, Node.js.\n\n```\n# On any project (empty or existing)\nnpx @patricksardinha/agentkit-cli init\n\n# With a blueprint\nnpx @patricksardinha/agentkit-cli init --blueprint PROJECT_BLUEPRINT.md\n\n# Add a feature to an existing project\nnpx @patricksardinha/agentkit-cli add --feature \"add dark mode\"\n```\n\nGitHub: [https://github.com/patricksardinha/agentkit-cli](https://github.com/patricksardinha/agentkit-cli)\n\nnpm: [https://www.npmjs.com/package/@patricksardinha/agentkit-cli](https://www.npmjs.com/package/@patricksardinha/agentkit-cli)\n\nFeedback welcome.", "url": "https://wpnews.pro/news/i-built-a-cli-that-scaffolds-agentic-workflows-for-claude-code", "canonical_source": "https://dev.to/patrick_sardinha/i-built-a-cli-that-scaffolds-agentic-workflows-for-claude-code-328c", "published_at": "2026-05-27 22:08:11+00:00", "updated_at": "2026-05-27 22:41:09.522371+00:00", "lang": "en", "topics": ["ai-agents", "ai-tools", "ai-products", "ai-startups", "large-language-models"], "entities": ["Claude Code", "AgentKit", "Next.js", "TypeScript", "Tailwind", "Patrick Sardinha"], "alternates": {"html": "https://wpnews.pro/news/i-built-a-cli-that-scaffolds-agentic-workflows-for-claude-code", "markdown": "https://wpnews.pro/news/i-built-a-cli-that-scaffolds-agentic-workflows-for-claude-code.md", "text": "https://wpnews.pro/news/i-built-a-cli-that-scaffolds-agentic-workflows-for-claude-code.txt", "jsonld": "https://wpnews.pro/news/i-built-a-cli-that-scaffolds-agentic-workflows-for-claude-code.jsonld"}}