{
  "markdown": "# conjure — Build Claude Code Extensions from Plain Language\n\n> Build Claude Code skills, agents, hooks, and MCP servers from plain-language descriptions. Describe what you want — conjure writes production-ready artifacts instantly. No boilerplate. No manual wiring.\n\n```\nclaude plugin marketplace add john-farina/claude-plugins\nclaude plugin install conjure@john-farina\n```\n\n**[→ conjure.dev landing page](https://john-farina.github.io/claude-plugins/)**\n\n---\n\n## What is conjure?\n\nConjure is a Claude Code plugin that builds the four types of Claude Code extensions:\n\n- **Skills** — slash commands you type (`/my-command`)\n- **Agents** — autonomous workers Claude spawns programmatically\n- **Hooks** — event-driven shell scripts that fire automatically\n- **MCP Servers** — live connections to external APIs, databases, and tools\n\nDescribe what you want in plain language. Conjure classifies the extension type, asks targeted questions, checks whether a published solution already exists, and writes a production-ready artifact. Every command also handles updates — no need to say \"update.\"\n\n---\n\n## How to build a Claude Code skill\n\n```\n/conjure-skill \"summarize my git changes and write a commit message in my team's style\"\n```\n\nConjure writes the `.md` command file with proper frontmatter, tool declarations, argument handling, and phased instructions. Immediately usable in the same session.\n\n---\n\n## How to build an MCP server for Claude Code\n\n```\n/conjure-mcp \"connect Claude to our internal Postgres database\"\n```\n\nConjure searches smithery.ai and mcp.so first. If no match exists, it scaffolds a full server in Node.js, Python, or Go with typed tools and registers it in `settings.json`.\n\n---\n\n## How to create a Claude Code agent\n\n```\n/conjure-agent \"review PRs and post feedback in my voice\"\n```\n\nConjure handles architecture decisions (orchestrator vs. worker), tool selection, `maxTurns`, `permissionMode`, and model config. Writes the definition to `~/.claude/agents/`.\n\n---\n\n## How to add a Claude Code hook\n\n```\n/conjure-hook \"run prettier on every file Claude edits\"\n```\n\nConjure writes the shell script and the `settings.json` entry. Supports all hook events: `PreToolUse`, `PostToolUse`, `SessionStart`, `UserPromptSubmit`, `Stop`.\n\n---\n\n## Commands\n\n| Command | What it does |\n|---|---|\n| `/conjure` | Auto-detects type from description and delegates |\n| `/conjure-skill` | Create or update a slash command / workflow |\n| `/conjure-agent` | Create or update an autonomous worker |\n| `/conjure-hook` | Create or update an event-driven hook |\n| `/conjure-mcp` | Create or update an MCP server |\n| `/conjure-config` | Set preferences all conjure commands follow automatically |\n| `/conjure-help` | Full reference and decision guide |\n\n---\n\n## conjure-config — preferences that apply everywhere\n\nRun `/conjure-config` once. Tell conjure your output paths, VCS tool, author name, naming conventions, required frontmatter — anything. It writes plain-language instructions to `~/.claude/conjure-config.md` (global) and `.claude/conjure-config.md` (project-local). Every conjure command reads these automatically and skips questions you've already answered.\n\n---\n\n## License\n\nMIT\n",
  "bytes": 3193,
  "sha": "fd48570174cbdf64e0ea5a6e89b9acb8b3359fe2d2fc319006c307a2ae466386",
  "repo_slug": "john-farina/claude-plugins",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/plg_john_farina_claude_plugins_conjure_30bc5cd6/readme"
}