@patricio0312rev/agentkit 0.1.0 → 0.2.0

package/CHANGELOG.md

@@ -0,0 +1,65 @@
+# Changelog
+
+All notable changes to AgentKit will be documented in this file.
+
+The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
+and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
+
+## [Unreleased]
+
+### Added
+- N/A
+
+### Changed
+- N/A
+
+### Fixed
+- N/A
+
+## [0.2.0] - 2025-12-28
+
+### Added
+- Model selection for Claude Code agents (sonnet, opus, haiku, or inherit from parent)
+- Interactive model selection prompt when choosing Claude Code tool
+- Automatic model field injection in agent frontmatter
+- Automated GitHub Actions workflow for npm publishing
+- `RELEASE.md` documentation for publishing process
+- `CHANGELOG.md` for tracking version changes
+
+### Changed
+- Claude Code folder structure: agents now placed directly in `.claude/agents` (no department subfolders)
+- Folder path validation now allows forward slashes for nested paths
+- Default folder for Claude Code changed from `.claude` to `.claude/agents`
+
+### Fixed
+- Folder validation rejecting valid paths with forward slashes
+- Model field not being added to agent frontmatter due to regex pattern mismatch
+
+## [0.1.0] - 2025-12-27
+
+### Added
+- Initial release of AgentKit
+- Support for Claude Code, Cursor, GitHub Copilot, Aider, and Universal AI tools
+- 7 department categories: Engineering, Design, Marketing, Product, Project Management, Studio Operations, Testing
+- 35+ pre-configured AI agent templates
+- Interactive CLI for agent selection
+- Department-based organization
+- Custom folder naming
+- Non-interactive mode with flags
+
+### Changed
+- N/A
+
+### Fixed
+- N/A
+
+---
+
+## Release Types
+
+- **Added** - New features
+- **Changed** - Changes in existing functionality
+- **Deprecated** - Soon-to-be removed features
+- **Removed** - Removed features
+- **Fixed** - Bug fixes
+- **Security** - Security fixes

package/README.md

@@ -4,7 +4,7 @@
 
 AgentKit is a CLI tool that helps you set up specialized AI agents for your development workflow. Get 42 pre-built agents across 7 departments—from engineering and design to marketing and testing—all optimized for rapid app development.
 
-[![npm version](https://img.shields.io/npm/v/agentkit.svg)](https://www.npmjs.com/package/agentkit)
+[![npm version](https://img.shields.io/npm/v/@patricio0312rev/agentkit.svg)](https://www.npmjs.com/package/@patricio0312rev/agentkit)
 [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
 
 ---
@@ -24,7 +24,7 @@ AgentKit is a CLI tool that helps you set up specialized AI agents for your deve
 
 ```bash
 # Install globally
-npm install -g agentkit
+npm install -g @patricio0312rev/agentkit
 
 # Or use directly with npx
 npx agentkit init
@@ -185,35 +185,6 @@ Copilot automatically uses `.github/copilot-instructions.md` for all suggestions
 
 ---
 
-## 🔧 Configuration Options
-
-```bash
-agentkit init [options]
-
-Options:
-  -t, --tool <tool>           AI tool (claude-code, cursor, copilot, aider, universal)
-  -f, --folder <name>         Custom folder name
-  -d, --departments <list>    Comma-separated list of departments
-  --agents <list>             Comma-separated list of specific agents
-  --skip-prompts              Skip all interactive prompts (requires all flags)
-  -h, --help                  Display help for command
-```
-
-**Examples:**
-
-```bash
-# Full engineering department
-agentkit init -t cursor -d engineering --skip-prompts
-
-# Specific agents only
-agentkit init -t cursor -d engineering --agents backend-architect,frontend-developer --skip-prompts
-
-# Multiple departments
-agentkit init -t claude-code -d engineering,design,marketing --skip-prompts
-```
-
----
-
 ## 📚 Documentation
 
 Each generated agent includes:
@@ -335,8 +306,8 @@ agentkit init \
 
 ```bash
 # If installed globally, try:
-npm uninstall -g agentkit
-npm install -g agentkit
+npm uninstall -g @patricio0312rev/agentkit
+npm install -g @patricio0312rev/agentkit
 
 # Or use npx:
 npx agentkit init
@@ -351,7 +322,7 @@ npx agentkit init
 **3. Permission errors during installation**
 
 ```bash
-sudo npm install -g agentkit
+sudo npm install -g @patricio0312rev/agentkit
 # Or use npx to avoid global install
 ```
 

package/RELEASE.md

@@ -0,0 +1,175 @@
+# Release Process
+
+This document explains how to publish new versions of AgentKit to npm using the automated GitHub Actions workflow.
+
+## One-Time Setup
+
+### 1. Get NPM Access Token
+
+1. Log in to [npmjs.com](https://www.npmjs.com/)
+2. Click on your profile icon → **Access Tokens**
+3. Click **Generate New Token** → **Classic Token**
+4. Select **Automation** type (allows CI/CD publishing)
+5. Copy the generated token (starts with `npm_...`)
+
+### 2. Add NPM Token to GitHub Secrets
+
+1. Go to your GitHub repository: `https://github.com/patricio0312rev/agentkit`
+2. Click **Settings** → **Secrets and variables** → **Actions**
+3. Click **New repository secret**
+4. Name: `NPM_TOKEN`
+5. Value: Paste your npm token
+6. Click **Add secret**
+
+That's it! The `GITHUB_TOKEN` is automatically provided by GitHub Actions.
+
+## Publishing a New Release
+
+### Step 1: Update Version
+
+Update the version in `package.json`:
+
+```bash
+# For patch release (0.1.0 → 0.1.1)
+npm version patch
+
+# For minor release (0.1.0 → 0.2.0)
+npm version minor
+
+# For major release (0.1.0 → 1.0.0)
+npm version major
+```
+
+This command will:
+- Update `package.json` version
+- Create a git commit with the version change
+- Create a git tag (e.g., `v0.1.1`)
+
+### Step 2: Update Changelog (Optional but Recommended)
+
+Create or update `CHANGELOG.md` with release notes:
+
+```markdown
+## [0.1.1] - 2025-01-XX
+
+### Added
+- Model selection for Claude Code agents
+- Direct agent placement in .claude/agents folder
+
+### Changed
+- Updated folder structure for Claude Code
+
+### Fixed
+- Bug fixes...
+```
+
+Commit the changelog:
+
+```bash
+git add CHANGELOG.md
+git commit -m "docs: update changelog for v0.1.1"
+```
+
+### Step 3: Push Tags to Trigger Release
+
+```bash
+# Push commits
+git push origin main
+
+# Push tags (this triggers the GitHub Action)
+git push origin --tags
+```
+
+### Step 4: Monitor the Release
+
+1. Go to **Actions** tab in GitHub repository
+2. You should see the "Publish to npm" workflow running
+3. The workflow will:
+   - Install dependencies
+   - Run tests
+   - Verify version matches tag
+   - Publish to npm
+   - Create GitHub release
+
+### Step 5: Verify Publication
+
+Check that the package was published:
+
+```bash
+npm view @patricio0312rev/agentkit version
+```
+
+Visit the npm page:
+https://www.npmjs.com/package/@patricio0312rev/agentkit
+
+## Manual Release (If Needed)
+
+If the automated workflow fails, you can publish manually:
+
+```bash
+# Make sure you're on the correct version
+npm version <patch|minor|major>
+
+# Login to npm
+npm login
+
+# Publish
+npm publish --access public
+```
+
+## Release Checklist
+
+Before releasing, ensure:
+
+- [ ] All tests pass: `npm test`
+- [ ] Code is linted and formatted
+- [ ] Version is bumped in `package.json`
+- [ ] `CHANGELOG.md` is updated
+- [ ] All changes are committed
+- [ ] Tags are pushed to GitHub
+
+## Version Numbering Guidelines
+
+Follow [Semantic Versioning](https://semver.org/):
+
+- **PATCH** (0.1.X) - Bug fixes, small improvements
+- **MINOR** (0.X.0) - New features, backward compatible
+- **MAJOR** (X.0.0) - Breaking changes
+
+## Troubleshooting
+
+### "Version already exists" error
+
+You tried to publish a version that already exists on npm. Bump the version again:
+
+```bash
+npm version patch
+git push origin --tags
+```
+
+### "Permission denied" error
+
+Your `NPM_TOKEN` might be invalid or expired:
+1. Generate a new token on npmjs.com
+2. Update the GitHub secret
+
+### Workflow doesn't trigger
+
+Make sure you pushed the tag:
+
+```bash
+git push origin --tags
+```
+
+The workflow only runs when tags matching `v*.*.*` are pushed.
+
+## Quick Reference
+
+```bash
+# Complete release process
+npm version patch              # Bump version
+git add CHANGELOG.md          # Add changelog
+git commit -m "docs: update changelog"
+git push origin main          # Push commits
+git push origin --tags        # Push tags → triggers workflow
+```

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@patricio0312rev/agentkit",
   "displayName": "AgentKit",
-  "version": "0.1.0",
+  "version": "0.2.0",
   "description": "CLI tool to scaffold AI agent configurations for Claude Code, Cursor, and other AI coding assistants",
   "main": "src/index.js",
   "bin": {

package/src/commands/init.js

@@ -50,7 +50,8 @@ function buildConfigFromFlags(options) {
     agents: options.agents
       ? options.agents.split(",").map((a) => a.trim())
       : [],
-    stack: [], // Removed for v0.1.0
+    model: options.model === "inherit" ? undefined : options.model,
+    stack: [],
   };
 }
 
@@ -91,8 +92,8 @@ async function promptUser(options) {
         validate: (input) => {
           if (!input.trim()) return "Folder name cannot be empty";
           if (input.includes(" ")) return "Folder name cannot contain spaces";
-          if (input.includes("\\") || input.includes("/"))
-            return "Folder name cannot contain path separators";
+          if (input.includes("\\"))
+            return "Folder name cannot contain backslashes";
           return true;
         },
       },
@@ -102,6 +103,46 @@ async function promptUser(options) {
     answers.folder = options.folder;
   }
 
+  // Step 2.5: Model selection for Claude Code
+  if (answers.tool === "claude-code" && !options.model) {
+    const modelChoices = [
+      {
+        name: `${chalk.bold("Sonnet")} ${chalk.gray("→")} ${chalk.dim("Balanced performance - best for most agents")}`,
+        value: "sonnet",
+        short: "Sonnet",
+      },
+      {
+        name: `${chalk.bold("Opus")} ${chalk.gray("→")} ${chalk.dim("Most capable for complex reasoning tasks")}`,
+        value: "opus",
+        short: "Opus",
+      },
+      {
+        name: `${chalk.bold("Haiku")} ${chalk.gray("→")} ${chalk.dim("Fast and efficient for simple tasks")}`,
+        value: "haiku",
+        short: "Haiku",
+      },
+      {
+        name: `${chalk.bold("Inherit from parent")} ${chalk.gray("→")} ${chalk.dim("Use the same model as the main conversation")}`,
+        value: "inherit",
+        short: "Inherit from parent",
+      },
+    ];
+
+    const modelAnswer = await inquirer.prompt([
+      {
+        type: "list",
+        name: "model",
+        message: "Select model for agents:",
+        choices: modelChoices,
+        default: "sonnet",
+      },
+    ]);
+    answers.model =
+      modelAnswer.model === "inherit" ? undefined : modelAnswer.model;
+  } else if (options.model) {
+    answers.model = options.model === "inherit" ? undefined : options.model;
+  }
+
   // Step 3: Select departments
   if (!options.departments) {
     const deptChoices = Object.entries(DEPARTMENTS).map(([key, value]) => ({

package/src/lib/config.js

@@ -1,7 +1,7 @@
 const TOOLS = {
     'claude-code': {
       name: 'Claude Code',
-      folder: '.claude',
+      folder: '.claude/agents',
       description: 'Sub-agents with native support',
       fileStructure: 'multi-file',
       supportsSubAgents: true

package/src/lib/generator.js

@@ -1,8 +1,8 @@
-const fs = require('fs-extra');
-const path = require('path');
-const { TOOLS, DEPARTMENTS } = require('./config');
-const { generateReadme } = require('../utils/readme');
-const { generateToolSpecificFiles } = require('../utils/tool-specific');
+const fs = require("fs-extra");
+const path = require("path");
+const { TOOLS, DEPARTMENTS } = require("./config");
+const { generateReadme } = require("../utils/readme");
+const { generateToolSpecificFiles } = require("../utils/tool-specific");
 
 /**
  * Main generator function that creates AI agent configuration
@@ -16,7 +16,7 @@ async function generateAgents(config) {
     targetDir,
     agentsGenerated: 0,
     filesCreated: [],
-    errors: []
+    errors: [],
   };
 
   try {
@@ -29,18 +29,21 @@ async function generateAgents(config) {
     result.filesCreated.push(...agentResult.files);
 
     // Generate README
-    const readmePath = path.join(targetDir, 'README.md');
+    const readmePath = path.join(targetDir, "README.md");
     const readmeContent = generateReadme(config, result);
     await fs.writeFile(readmePath, readmeContent);
     result.filesCreated.push(readmePath);
 
     // Generate tool-specific files (e.g., .cursorrules, copilot-instructions.md)
-    const toolFiles = await generateToolSpecificFiles(config, targetDir, agentResult.agentsList);
+    const toolFiles = await generateToolSpecificFiles(
+      config,
+      targetDir,
+      agentResult.agentsList
+    );
     result.filesCreated.push(...toolFiles);
 
     result.success = true;
     return result;
-
   } catch (error) {
     result.errors.push(error.message);
     throw error;
@@ -58,22 +61,28 @@ async function generateDepartmentAgents(config, targetDir) {
   const files = [];
   const agentsList = [];
 
+  // For Claude Code, agents go directly in .claude/agents, not in department subfolders
+  const isClaudeCode = config.tool === "claude-code";
+
   for (const dept of config.departments) {
     const deptInfo = DEPARTMENTS[dept];
     if (!deptInfo) continue;
 
-    const deptDir = path.join(targetDir, dept);
+    // Only create department subdirectories for non-Claude Code tools
+    const deptDir = isClaudeCode ? targetDir : path.join(targetDir, dept);
     await fs.ensureDir(deptDir);
 
     // Determine which agents to generate
     let agentsToGenerate = deptInfo.agents;
-    
+
     // Filter if specific agents were selected
     if (config.agents && config.agents.length > 0) {
-      agentsToGenerate = agentsToGenerate.filter(agent => {
+      agentsToGenerate = agentsToGenerate.filter((agent) => {
         // Support both "agent-name" and "dept/agent-name" formats
-        return config.agents.includes(agent) || 
-               config.agents.includes(`${dept}/${agent}`);
+        return (
+          config.agents.includes(agent) ||
+          config.agents.includes(`${dept}/${agent}`)
+        );
       });
     }
 
@@ -100,13 +109,18 @@ async function generateDepartmentAgents(config, targetDir) {
  * @returns {Promise<string|null>} Path to created file or null
  */
 async function generateAgentFile(dept, agent, deptDir, config) {
-  const sourceFile = path.join(__dirname, '../../templates/departments', dept, `${agent}.md`);
+  const sourceFile = path.join(
+    __dirname,
+    "../../templates/departments",
+    dept,
+    `${agent}.md`
+  );
   const targetFile = path.join(deptDir, `${agent}.md`);
 
   try {
     // Check if template exists
     if (await fs.pathExists(sourceFile)) {
-      let content = await fs.readFile(sourceFile, 'utf8');
+      let content = await fs.readFile(sourceFile, "utf8");
 
       // Process content based on config
       content = processAgentContent(content, config);
@@ -115,7 +129,7 @@ async function generateAgentFile(dept, agent, deptDir, config) {
       return targetFile;
     } else {
       // Generate basic template if source doesn't exist
-      const basicTemplate = generateBasicAgentTemplate(dept, agent);
+      const basicTemplate = generateBasicAgentTemplate(dept, agent, config);
       await fs.writeFile(targetFile, basicTemplate);
       return targetFile;
     }
@@ -134,17 +148,34 @@ async function generateAgentFile(dept, agent, deptDir, config) {
 function processAgentContent(content, config) {
   let processed = content;
 
+  // Add model field to frontmatter for Claude Code agents
+  if (config.tool === "claude-code" && config.model) {
+    const frontmatterMatch = processed.match(/^---\n([\s\S]*?)\n---/);
+    if (frontmatterMatch) {
+      const frontmatterContent = frontmatterMatch[1];
+
+      const updatedFrontmatter = frontmatterContent.replace(
+        /(tools:.*?)($|\n)/,
+        `$1\nmodel: ${config.model}$2`
+      );
+      processed = processed.replace(
+        /^---\n[\s\S]*?\n---/,
+        `---\n${updatedFrontmatter}\n---`
+      );
+    }
+  }
+
   // Remove examples if requested
   if (config.skipExamples) {
     // Remove content between <example> tags
-    processed = processed.replace(/<example>[\s\S]*?<\/example>/g, '');
+    processed = processed.replace(/<example>[\s\S]*?<\/example>/g, "");
     // Clean up extra whitespace
-    processed = processed.replace(/\n{3,}/g, '\n\n');
+    processed = processed.replace(/\n{3,}/g, "\n\n");
   }
 
   // Add tech stack information if provided
   if (config.stack && config.stack.length > 0) {
-    const stackSection = `\n\n## Tech Stack Context\n\nThis project uses: ${config.stack.join(', ')}\n`;
+    const stackSection = `\n\n## Tech Stack Context\n\nThis project uses: ${config.stack.join(", ")}\n`;
     // Insert before the main content or at the end
     processed += stackSection;
   }
@@ -156,19 +187,27 @@ function processAgentContent(content, config) {
  * Generate a basic agent template when source file doesn't exist
  * @param {string} dept - Department name
  * @param {string} agent - Agent name
+ * @param {Object} config - Configuration object
  * @returns {string} Basic template content
  */
-function generateBasicAgentTemplate(dept, agent) {
+function generateBasicAgentTemplate(dept, agent, config) {
   const deptInfo = DEPARTMENTS[dept];
+  const modelLine =
+    config.tool === "claude-code" && config.model
+      ? `\nmodel: ${config.model}`
+      : "";
   return `<!-- ${dept}/${agent}.md -->
 ---
 name: ${agent}
 description: AI agent for ${deptInfo.name}
 color: blue
-tools: Read, Write, MultiEdit
+tools: Read, Write, MultiEdit${modelLine}
 ---
 
-# ${agent.split('-').map(w => w.charAt(0).toUpperCase() + w.slice(1)).join(' ')}
+# ${agent
+    .split("-")
+    .map((w) => w.charAt(0).toUpperCase() + w.slice(1))
+    .join(" ")}
 
 This is a placeholder agent file. Please add specific instructions and responsibilities.
 
@@ -189,5 +228,5 @@ This is a placeholder agent file. Please add specific instructions and responsib
 module.exports = {
   generateAgents,
   generateDepartmentAgents,
-  generateAgentFile
+  generateAgentFile,
 };