claude-setup 1.1.3 → 1.1.4

package/package.json

@@ -1,49 +1,49 @@
-{
-  "name": "claude-setup",
-  "version": "1.1.3",
-  "description": "Setup layer for Claude Code — reads your project, writes command files, Claude Code does the rest",
-  "type": "module",
-  "bin": {
-    "claude-setup": "./dist/index.js"
-  },
-  "scripts": {
-    "build": "tsc",
-    "dev": "tsc --watch",
-    "prepublishOnly": "tsc"
-  },
-  "files": [
-    "dist",
-    "templates"
-  ],
-  "keywords": [
-    "claude",
-    "claude-code",
-    "cli",
-    "developer-tools",
-    "ai",
-    "anthropic",
-    "automation",
-    "scaffolding",
-    "project-setup",
-    "mcp",
-    "npx"
-  ],
-  "author": "AbdoKnbGit",
-  "license": "MIT",
-  "repository": {
-    "type": "git",
-    "url": "git+https://github.com/AbdoKnbGit/claude-setup.git"
-  },
-  "homepage": "https://github.com/AbdoKnbGit/claude-setup#readme",
-  "bugs": {
-    "url": "https://github.com/AbdoKnbGit/claude-setup/issues"
-  },
-  "dependencies": {
-    "commander": "^12.1.0",
-    "glob": "^11.0.0"
-  },
-  "devDependencies": {
-    "@types/node": "^22.0.0",
-    "typescript": "^5.6.0"
-  }
-}
+{
+  "name": "claude-setup",
+  "version": "1.1.4",
+  "description": "Setup layer for Claude Code — reads your project, writes command files, Claude Code does the rest",
+  "type": "module",
+  "bin": {
+    "claude-setup": "./dist/index.js"
+  },
+  "scripts": {
+    "build": "tsc",
+    "dev": "tsc --watch",
+    "prepublishOnly": "tsc"
+  },
+  "files": [
+    "dist",
+    "templates"
+  ],
+  "keywords": [
+    "claude",
+    "claude-code",
+    "cli",
+    "developer-tools",
+    "ai",
+    "anthropic",
+    "automation",
+    "scaffolding",
+    "project-setup",
+    "mcp",
+    "npx"
+  ],
+  "author": "AbdoKnbGit",
+  "license": "MIT",
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/AbdoKnbGit/claude-setup.git"
+  },
+  "homepage": "https://github.com/AbdoKnbGit/claude-setup#readme",
+  "bugs": {
+    "url": "https://github.com/AbdoKnbGit/claude-setup/issues"
+  },
+  "dependencies": {
+    "commander": "^12.1.0",
+    "glob": "^11.0.0"
+  },
+  "devDependencies": {
+    "@types/node": "^22.0.0",
+    "typescript": "^5.6.0"
+  }
+}

package/templates/add.md

@@ -24,14 +24,133 @@ Add to Claude Code setup: "{{USER_INPUT}}"
 
 Skills: {{SKILLS_LIST}} | Commands: {{COMMANDS_LIST}}
 
+---
+
+{{MARKETPLACE_INSTRUCTIONS}}
+
+---
+
+## What you must actually do
+
+Parse the user's request and take ALL applicable actions:
+
+### 1. MCP servers
+If the request mentions an external service (database, API, browser, etc.):
+- Check the verified MCP package list below
+- If found: add to `.mcp.json` with OS-correct format (detected: {{DETECTED_OS}})
+- Use `${VARNAME}` syntax for all credentials — NEVER hardcode
+- Document new env vars in `.env.example`
+
+Verified MCP packages — ONLY use these for MCP servers:
+```
+playwright   → @playwright/mcp@latest
+postgres     → @modelcontextprotocol/server-postgres
+filesystem   → @modelcontextprotocol/server-filesystem
+memory       → @modelcontextprotocol/server-memory
+github       → @modelcontextprotocol/server-github
+brave        → @modelcontextprotocol/server-brave-search
+puppeteer    → @modelcontextprotocol/server-puppeteer
+slack        → @modelcontextprotocol/server-slack
+sqlite       → @modelcontextprotocol/server-sqlite
+stripe       → @stripe/mcp@latest
+redis        → @modelcontextprotocol/server-redis
+mysql        → @benborla29/mcp-server-mysql
+mongodb      → mcp-mongo-server
+```
+
+{{#if HAS_MCP_JSON}}
+MCP format — merge into existing .mcp.json:
+{{else}}
+MCP format — create new .mcp.json:
+{{/if}}
+{{#if IS_WINDOWS}}
+```json
+{
+  "mcpServers": {
+    "server-name": {
+      "command": "cmd",
+      "args": ["/c", "npx", "-y", "<package>"],
+      "env": {
+        "API_KEY": "${API_KEY}"
+      }
+    }
+  }
+}
+```
+{{else}}
+```json
+{
+  "mcpServers": {
+    "server-name": {
+      "command": "npx",
+      "args": ["-y", "<package>"],
+      "env": {
+        "API_KEY": "${API_KEY}"
+      }
+    }
+  }
+}
+```
+{{/if}}
+
+### 2. Skills
+If the request mentions skills or capabilities:
+- Create `.claude/skills/<name>/SKILL.md` with proper frontmatter
+- Use `description:` so Claude knows when to load the skill
+- Search the marketplace for matching pre-built skills (see above)
+
+Skill format:
+```yaml
+---
+name: skill-name
+description: What this skill does
+---
+
+Instructions...
+```
+
+### 3. Hooks
+If the request implies automated actions (formatting, building, notifications):
+- Add to `.claude/settings.json` using the CORRECT hooks format
+- Verify the tool exists before adding a hook for it
+
+Correct hooks format:
+```json
+{
+  "hooks": {
+    "PostToolUse": [
+      {
+        "matcher": "Edit|Write",
+        "hooks": [
+          {
+            "type": "command",
+            "command": "<shell command>"
+          }
+        ]
+      }
+    ]
+  }
+}
+```
+
+### 4. Plugins
+If the request matches a marketplace category or SaaS platform:
+- Suggest the relevant plugin with install commands
+- Show the user exactly how to install it
+
+### 5. CLAUDE.md
+Document any new capabilities, services, or patterns added.
+
 ## Rules
 - Read current content above before writing. Merge/append only.
 - If request mentions something not evidenced in project files: ask first.
-- OS detected: {{DETECTED_OS}}. Use correct command format for MCP/hooks:
-  - Windows: `{ "command": "cmd", "args": ["/c", "npx", "<pkg>"] }`
-  - macOS/Linux: `{ "command": "npx", "args": ["<pkg>"] }`
+- OS detected: {{DETECTED_OS}}. Use correct command format for MCP/hooks.
 - All env var refs use `${VARNAME}` syntax. Document new vars in .env.example.
+- **NEVER write a "model" key into settings.json**
+- Produce valid JSON only.
 
 ## Output — one line per file
 Updated: ✅ [path] — [what and why]
+Created: ✅ [path] — [what and why]
 Skipped: ⏭ [path] — [why not needed for this request]
+Suggested: 📦 [plugin name] — install with: /plugin install [name]@[marketplace]

package/templates/init-empty.md

@@ -24,12 +24,69 @@ If they gave no language: write a language-agnostic CLAUDE.md about the product
 **.mcp.json**
 Only for services they explicitly mentioned or obviously required for their product type.
 If they said "not sure": do not create this file.
+If they mentioned a database, payment system, or external API, create .mcp.json.
+
+Use OS-correct format (detected: {{DETECTED_OS}}):
+```json
+{
+  "mcpServers": {
+    "server-name": {
+      "command": "npx",
+      "args": ["-y", "@package/name"],
+      "env": {
+        "API_KEY": "${API_KEY}"
+      }
+    }
+  }
+}
+```
+Windows format: use `"command": "cmd", "args": ["/c", "npx", "-y", "@package/name"]`
+
+Verified MCP packages:
+- postgres → @modelcontextprotocol/server-postgres
+- mysql → @benborla29/mcp-server-mysql
+- mongodb → mcp-mongo-server
+- redis → @modelcontextprotocol/server-redis
+- stripe → @stripe/mcp@latest
+- github → @modelcontextprotocol/server-github
+- filesystem → @modelcontextprotocol/server-filesystem
+- memory → @modelcontextprotocol/server-memory
+- playwright → @playwright/mcp@latest
 
 **.claude/settings.json**
 Only if the product type they described has clear patterns that benefit from hooks.
+Use the CORRECT hooks format:
+```json
+{
+  "hooks": {
+    "PostToolUse": [
+      {
+        "matcher": "Edit|Write",
+        "hooks": [
+          {
+            "type": "command",
+            "command": "<shell command>"
+          }
+        ]
+      }
+    ]
+  }
+}
+```
+**NEVER write a "model" key** — it overrides the user's model selection.
+**Before adding a build hook**, verify the tool is installed first.
 
 **.claude/skills/**
-Only if the product domain has patterns worth capturing. Empty is fine for a fresh start.
+Only if the product domain has patterns worth capturing. Each skill needs:
+```
+.claude/skills/<name>/SKILL.md
+```
+With frontmatter: `---\nname: ...\ndescription: ...\n---`
+Empty is fine for a fresh start.
+
+**.claude/commands/**
+Create project-appropriate commands based on what the user described.
+For example: /dev, /build, /test, /deploy — but only if the stack warrants them.
 
 **.github/workflows/**
 Only if they mentioned CI, deployment, or a specific hosting platform.

package/templates/init.md

@@ -32,20 +32,15 @@ CLAUDE.md → does not exist (create it)
 {{MCP_JSON_CONTENT}}
 
 OS detected: {{DETECTED_OS}}. Use correct MCP command format:
-- Windows: `{ "command": "cmd", "args": ["/c", "npx", "<package>"] }`
-- macOS/Linux: `{ "command": "npx", "args": ["<package>"] }`
+- Windows: `{ "command": "cmd", "args": ["/c", "npx", "-y", "<package>"] }`
+- macOS/Linux: `{ "command": "npx", "args": ["-y", "<package>"] }`
 {{else}}
-.mcp.json → does not exist (create only if you find evidence of external services)
+.mcp.json → does not exist (create if you find evidence of external services in deps, docker-compose, or env vars)
 {{/if}}
 
 {{#if HAS_SETTINGS}}
 ### .claude/settings.json — EXISTS — merge only, never remove existing hooks
 {{SETTINGS_CONTENT}}
-
-Hook shell format for {{DETECTED_OS}}:
-- Windows: `{ "command": "cmd", "args": ["/c", "<command>"] }`
-- macOS/Linux: `{ "command": "bash", "args": ["-c", "<command>"] }`
-- Bash quoting rule: never use bare `"` inside `-c "..."` — use `\x22` instead
 {{else}}
 settings.json → does not exist (create only if hooks are warranted)
 {{/if}}
@@ -70,21 +65,63 @@ actual conventions from the code. Generic advice belongs in docs, not here.
 If it exists: read it above first. Add only what is genuinely missing. Never remove.
 
 ### .mcp.json
-Only if you found evidence of external services in the config files, dependencies,
-or environment template. No evidence = no server.
+Create if you found evidence of external services in: dependencies, docker-compose services,
+env vars (DATABASE_URL, REDIS_URL, STRIPE_KEY, etc.), or import statements (pg, mysql2, mongoose, redis, stripe).
 If it exists: add to it. Never remove existing entries. Produce valid JSON.
-Use OS-correct command format (see above).
+Use OS-correct command format (see above). Always include `-y` in npx args.
+All credentials use `${VARNAME}` syntax — NEVER hardcode connection strings.
 
-### .claude/settings.json
+### .claude/settings.json — CORRECT HOOKS FORMAT
 Only if hooks genuinely earn their cost for this specific project.
 Every hook adds overhead on every Claude Code action.
+
+**Use this exact format:**
+```json
+{
+  "hooks": {
+    "PostToolUse": [
+      {
+        "matcher": "Edit|Write",
+        "hooks": [
+          {
+            "type": "command",
+            "command": "<shell command here>"
+          }
+        ]
+      }
+    ]
+  }
+}
+```
+
+**Before adding a build hook**, verify the tool is installed:
+- Wrap with existence check: `command -v mvn && mvn compile -q`
+- If not installed: skip the hook and warn the user
+- NEVER add a hook for a tool that doesn't exist
+
+**NEVER write a `"model"` key** — it overrides the user's model selection.
+
 If it exists: add to it. Never remove existing hooks.
-Use OS-correct shell format (see above).
+Use OS-correct shell format.
 
 ### .claude/skills/
-Only for patterns that recur across this codebase and benefit from automatic loading.
-Use `applies-when` frontmatter so skills load only when relevant.
-If a similar skill already exists: extend it.
+Create skills for patterns that recur across this codebase and benefit from automatic loading.
+Each skill must be a directory with SKILL.md:
+```
+.claude/skills/<name>/SKILL.md
+```
+With frontmatter:
+```yaml
+---
+name: skill-name
+description: When to use this skill
+---
+Instructions...
+```
+
+### .claude/commands/
+Create project-specific commands for multi-step workflows developers repeat.
+Based on what you find in scripts, Makefile, README, docker-compose.
 
 ### .github/workflows/
 Only if .github/ exists ({{HAS_GITHUB_DIR}}).

package/templates/remove.md

@@ -30,8 +30,32 @@ Before removing anything, scan these locations directly:
 - Skills   : `.claude/skills/*/SKILL.md`, `.claude/skills/*.md`, `.claude/skills/**/*.md`
 - Commands : `.claude/commands/*.md` (exclude stack-*.md)
 - MCP      : read `.mcp.json` directly
-- Hooks    : read `.claude/settings.json` directly
+- Hooks    : read `.claude/settings.json` → look inside the `"hooks"` key
 - CLAUDE.md: read the file directly
+- Plugins  : check `/plugin` installed list
+
+## What to remove for each type
+
+### MCP servers
+- Remove the server entry from `.mcp.json` → `mcpServers.<name>`
+- Remove corresponding env vars from `.env.example` (if no other server uses them)
+- Remove references in CLAUDE.md
+
+### Hooks
+- Remove from `.claude/settings.json` → `hooks.<EventName>` entries
+- Hooks use this structure: `{ "hooks": { "PostToolUse": [{ "matcher": "...", "hooks": [...] }] } }`
+- Remove the entire matcher entry if removing all hooks for that matcher
+
+### Skills
+- Delete the skill directory: `.claude/skills/<name>/` (entire directory including SKILL.md)
+- Or delete the flat skill file: `.claude/skills/<name>.md`
+
+### Plugins
+- Suggest: `/plugin uninstall <name>@<marketplace>`
+- Print the exact uninstall command for the user
+
+### Commands
+- Delete the command file: `.claude/commands/<name>.md`
 
 Before deleting, list every reference found and confirm scope:
 ```
@@ -45,7 +69,7 @@ Dangling references that will break:
 ## Rules
 - Find everything related to the removal request across ALL files — scan the filesystem, not just the data shown above.
 - Remove surgically — section by section, key by key.
-- Never delete an entire file. Remove only the relevant section.
+- Never delete an entire file unless it contains ONLY the thing being removed.
 - Never remove content unrelated to the request.
 - After every edit, the file MUST remain valid (JSON stays valid JSON, etc.)
 - If not found anywhere: say so and stop.
@@ -55,3 +79,4 @@ Dangling references that will break:
 Removed: ✅ [path] — [what was removed]
 Not found: ⏭ [path] — not referenced
 Dangling: ⚠️ [path] — still references [removed thing]
+Suggested: 📦 To uninstall plugin: /plugin uninstall [name]@[marketplace]

package/templates/sync.md

@@ -30,7 +30,7 @@ Project changed since last setup. Update ONLY what the changes demand.
 {{SETTINGS_CONTENT}}
 {{/if}}
 
-Skills: {{SKILLS_LIST}} | Commands: {{COMMANDS_LIST}}
+Skills: {{SKILLS_LIST}} | Commands: {{COMMANDS_LIST}} | Workflows: {{WORKFLOWS_LIST}}
 
 ## Your job
 
@@ -47,6 +47,25 @@ Do NOT update things that did not change.
 Do NOT rewrite files — surgical edits only.
 If unsure about a change's implication: flag it, don't guess.
 
+### Correct hooks format (if adding/modifying hooks)
+```json
+{
+  "hooks": {
+    "PostToolUse": [
+      {
+        "matcher": "Edit|Write",
+        "hooks": [
+          {
+            "type": "command",
+            "command": "<shell command>"
+          }
+        ]
+      }
+    ]
+  }
+}
+```
+
 ## Output — one line per file, nothing else
 
 Updated: ✅ [path] — triggered by: [which changed file and why]