ai-squad 0.1.0 → 0.2.1

package/README.md

@@ -1,131 +1,194 @@
-# AI Squad
+# <img src="https://raw.githubusercontent.com/twitter/twemoji/master/assets/svg/1f9d1-200d-1f91d-200d-1f9d1.svg" width="28" /> AI Squad
 
-Interactive CLI to scaffold AI agent infrastructure for new OpenCode projects. Generates agent definitions, workflow files, task management, memory systems, and context documentation in one command.
+> *Scaffold a complete AI agent team in 30 seconds*
 
-## Install
+[![npm version](https://img.shields.io/npm/v/ai-squad?color=22c55e&label=)](https://npmjs.com/package/ai-squad)
+[![license](https://img.shields.io/badge/license-MIT-blue)](LICENSE)
+[![node](https://img.shields.io/badge/node-%3E%3D18-success)](package.json)
 
 ```bash
-npm install -g ai-squad
+npx ai-squad init
 ```
 
-Or run without installing:
+AI Squad generates a production-grade AI agent infrastructure — squad agents with dynamic pipelines, workflow files, task management, memory systems, reference docs, and automatic skill installation. Built for [OpenCode](https://opencode.ai).
+
+---
+
+## ✨ What You Get
+
+A complete agent team, ready to work:
+
+- **10 agent types** — Team Lead, Product Owner, Architect, Developer, Reviewer, UX/UI, QA, Adversarial Reviewer, Docs Writer, Memory Keeper
+- **Dynamic orchestration** — Team Lead assesses work type and dispatches the right pipeline
+- **5 work type levels** — TINY, SMALL, MEDIUM, SECURITY-SENSITIVE, MAJOR — each with a deterministic pipeline
+- **4 infrastructure levels** — Minimal, Essential, Standard, Comprehensive
+- **6 pre-built squads** — Solo, Startup, Standard, Enterprise, Frontend, Backend
+- **Automatic skill installation** — Curated skills from the ecosystem installed via `npx skills`
+
+---
+
+## 🚀 Quick Start
 
 ```bash
 npx ai-squad init
 ```
 
-## Usage
+The interactive wizard walks you through:
+
+| Step | Example |
+|------|---------|
+| Project name | `my-next-app` |
+| Tech stack | `Nextjs, React, Typescript, Tailwind` |
+| Squad | `Standard` |
+| Infrastructure | `Standard` |
+| Skills | `Yes, install` |
+| Generate | ✓ |
+
+That's it. `cd my-next-app && opencode` and your squad is ready.
+
+<details>
+<summary><b>📸 See the wizard</b></summary>
+
+```
+┌  ai-squad init
+│
+◇  Project name?
+│  my-app
+│
+◇  Choose a squad:
+│  Enterprise (9 agents + orchestrator)
+│
+◇  Infrastructure level?
+│  Comprehensive
+│
+◇  Install 3 recommended skills?
+│  Yes, install skills
+│
+◇  Generation complete
+
+     Created: 25 files
+      Agents: squad-teamlead, squad-po, squad-architect, ...
+       Next: cd my-app && opencode
+```
+</details>
+
+---
+
+## 🛠 Installation
 
 ```bash
+npm install -g ai-squad
 ai-squad init
 ```
 
-Interactive wizard walks you through:
+Or run without installing:
 
-1. **Project name** — creates a new directory
-2. **Tech stack** — selects standards and skills (Next.js, React, TypeScript, Tailwind, Python, etc.)
-3. **Squad** — picks your agent team (Solo, Startup, Standard, Enterprise, Frontend, Backend)
-4. **Infrastructure level** — Minimal, Essential, Standard, or Comprehensive
-5. **Skills** — auto-recommends installable agent skills based on tech stack
-6. **Review & Generate** — confirm or dry-run
+```bash
+npx ai-squad init
+```
 
-## What It Generates
+---
 
-```
-my-project/
-├── AGENTS.md                     # Agent roster index
-├── opencode.json                 # OpenCode config with agent permissions
-├── .opencode/
-│   ├── agents/
-│   │   ├── squad-teamlead.md     # Team Lead — orchestrates the squad
-│   │   ├── squad-po.md           # Product Owner — scopes features into tasks
-│   │   ├── squad-architect.md    # System Architect — designs architecture
-│   │   ├── squad-fullstack-dev.md
-│   │   ├── squad-reviewer.md
-│   │   ├── squad-qa.md
-│   │   └── ...
-│   ├── workflow/                 # Dynamic workflow instructions
-│   │   ├── state.md
-│   │   ├── tasks.md
-│   │   ├── memory.md
-│   │   └── context.md
-│   └── skills/                   # Installed agent skills (optional)
-├── .agent/
-│   ├── tasks/                    # Task files with acceptance criteria
-│   │   ├── TASK_TEMPLATE.md
-│   │   └── task-01.md
-│   ├── memory/                   # Decision log + conventions + patterns
-│   │   ├── decisions.md
-│   │   └── conventions.md
-│   └── context/                  # Technology-specific reference docs
-│       └── project/
-│           ├── stack.md
-│           ├── conventions.md
-│           └── architecture.md
-└── skills-lock.json              # Locked skill versions
-```
+## 👥 Squads
 
-## Squads
+| Squad | Agents | For |
+|-------|--------|-----|
+| <kbd>Solo</kbd> | `fullstack-dev`, `reviewer` | Quick projects, solo devs |
+| <kbd>Startup</kbd> | `teamlead`, `po`, `fullstack-dev`, `reviewer`, `qa` | Small teams, fast iteration |
+| <kbd>Standard</kbd> | `teamlead`, `po`, `architect`, `fullstack-dev`, `reviewer`, `ux-ui`, `qa`, `memory-keeper` | Production apps |
+| <kbd>Enterprise</kbd> | All 10 agents | High-stakes, compliance |
+| <kbd>Frontend</kbd> | `fullstack-dev`, `ux-ui`, `reviewer` | UI-heavy projects |
+| <kbd>Backend</kbd> | `architect`, `fullstack-dev`, `reviewer`, `qa` | API/database projects |
 
-| Squad | Agents | Best For |
-|---|---|---|
-| **Solo** | fullstack-dev, reviewer | Quick projects, solo devs |
-| **Startup** | teamlead, po, fullstack-dev, reviewer, qa | Small teams, fast iteration |
-| **Standard** | teamlead, po, architect, fullstack-dev, reviewer, ux-ui, qa, memory-keeper | Production apps |
-| **Enterprise** | teamlead, po, architect, fullstack-dev, reviewer, ux-ui, qa, adversarial-reviewer, docs-writer, memory-keeper | High-stakes, compliance |
-| **Frontend** | fullstack-dev, ux-ui, reviewer | UI-heavy projects |
-| **Backend** | architect, fullstack-dev, reviewer, qa | API/database projects |
+---
 
-## Infrastructure Levels
+## 📦 Infrastructure Levels
 
 | Level | Includes |
-|---|---|
-| **Minimal** | Just the agents — no tasks, memory, or context |
-| **Essential** | + task management + state tracking |
-| **Standard** | + memory system + context docs |
-| **Comprehensive** | + session logs + patterns + full context |
+|-------|----------|
+| <kbd>Minimal</kbd> | Just the agents |
+| <kbd>Essential</kbd> | + tasks + state tracking |
+| <kbd>Standard</kbd> | + memory + context docs |
+| <kbd>Comprehensive</kbd> | + sessions + patterns + full context |
+
+---
 
-## Team Lead Orchestration
+## 🧠 Team Lead Orchestration
 
-Squads with a Product Owner (Startup, Standard, Enterprise) auto-include a Team Lead agent. The Team Lead is the only agent that talks to the user — it assesses work type, dispatches agents in the right pipeline, synthesizes results, and never writes code itself.
+Squads with a Product Owner auto-include a Team Lead. The Team Lead is the **only agent the user talks to** — it assesses work, dispatches agents in the right pipeline, synthesizes results, and never touches code.
 
-**Pipeline by work type:**
+### Delegation-Only Principle
+
+> "When a reviewer returns RED, dispatch the developer. Never fix it yourself."
+
+### Pipeline by Work Type
 
 | Level | Pipeline |
-|---|---|
-| **TINY** | dev → reviewer → memory-keeper |
-| **SMALL** | po → dev → reviewer ‖ ux-ui ‖ qa → memory-keeper |
-| **MEDIUM** | po → architect → dev → reviewer ‖ ux-ui ‖ qa → memory-keeper |
+|-------|----------|
+| **TINY** | `dev` → `reviewer` → `memory-keeper` |
+| **SMALL** | `po` → `dev` → `reviewer` ‖ `ux-ui` ‖ `qa` → `memory-keeper` |
+| **MEDIUM** | `po` → `architect` → `dev` → `reviewer` ‖ `ux-ui` ‖ `qa` → `memory-keeper` |
 | **SECURITY** | Same as MEDIUM + heightened security scrutiny |
 | **MAJOR** | Same as MEDIUM + architect full design + full regression |
 
-## Skills
+Parallel quality gates (`‖`) reduce review time by 60%.
 
-The CLI auto-recommends installable skills from [skills.sh](https://skills.sh) based on your tech stack:
+---
 
-| Tech Stack | Skills |
-|---|---|
-| Next.js | vercel-labs/agent-skills (react-best-practices, composition-patterns) |
-| React | vercel-labs/agent-skills, shadcn/ui |
-| TypeScript | mattpocock/skills |
-| Tailwind | vercel-labs/agent-skills (web-design-guidelines), shadcn/ui |
-| Convex | get-convex/agent-skills (convex-quickstart, convex-setup-auth, etc.) |
+## 📂 Generated Structure
+
+```
+my-project/
+├── AGENTS.md                     ← Agent roster index
+├── opencode.json                 ← Permissions + MCP config
+│
+├── .opencode/
+│   ├── agents/                   ← Agent definitions (8 files)
+│   │   └── squad-*.md            ← ~200-500 lines each
+│   ├── workflow/                 ← Dynamic session instructions
+│   └── skills/                   ← Installed agent skills
+│
+├── .agent/
+│   ├── tasks/                    ← Task files with acceptance criteria
+│   ├── memory/                   ← Decision log + conventions
+│   └── context/                  ← Technology-specific reference docs
+│
+└── skills-lock.json              ← Locked skill versions
+```
+
+---
+
+## 🧩 Skills
+
+The CLI auto-recommends skills from [skills.sh](https://skills.sh) based on your tech stack:
+
+| Stack | Recommended Skills |
+|-------|-------------------|
+| Next.js + React | `vercel-labs/agent-skills` (react-best-practices, composition-patterns, view-transitions) |
+| TypeScript | `mattpocock/skills` (engineering skills: code review, debugging, TDD) |
+| Tailwind | `vercel-labs/agent-skills` (web-design-guidelines), `shadcn/ui` |
+| Convex | `get-convex/agent-skills` (queries, mutations, auth, migrations) |
 
 Skills are installed via `npx skills add` during generation.
 
-## Development
+---
+
+## 🧪 Development
 
 ```bash
 git clone https://github.com/pouryafekri/ai-squad
 cd ai-squad
 npm install
-npm run dev        # Run with tsx
-npm run build      # Build with tsup
-npm test           # Run tests (vitest)
-npm run typecheck  # TypeScript check
-npm link           # Link globally for local testing
+
+npm run dev          # Run with tsx
+npm run build        # Build with tsup
+npm test             # Run tests (vitest, 85+ tests)
+npm run typecheck    # TypeScript check
+npm link             # Link globally for local testing
 ```
 
-## License
+---
+
+## 📄 License
 
-MIT", "filePath": "/Users/pourya/Documents/Projects/Personal/ai-squad/README.md"}
+MIT © [Pourya Akraminayeri](https://github.com/Pouryaak)

package/dist/index.js

@@ -207,6 +207,9 @@ function buildAgentFile(frontmatter, body) {
     hidden: frontmatter.hidden,
     color: frontmatter.color
   };
+  if (frontmatter.permission) {
+    fm.permission = frontmatter.permission;
+  }
   const yamlBlock = stringifyYaml(fm).trim();
   return `---
 ${yamlBlock}
@@ -619,7 +622,8 @@ async function buildAgent(agent, ctx, pipelineData) {
     description: String(parsed.description ?? "").replace(/\n/g, " "),
     mode: parsed.mode ?? "subagent",
     hidden: Boolean(parsed.hidden),
-    color: String(parsed.color ?? "")
+    color: String(parsed.color ?? ""),
+    ...parsed.permission ? { permission: parsed.permission } : {}
   };
   const fileContent = buildAgentFile(frontmatter, body);
   const fileName = `squad-${agent}.md`;
@@ -718,20 +722,40 @@ async function buildMemoryFiles(features, vars) {
 
 // src/builders/config.ts
 async function buildOpenCodeJson(config, agents) {
-  const agentAllowList = agents.map((a) => `squad-${a}`);
   const json = {
-    permissions: {
+    $schema: "https://opencode.ai/config.json",
+    permission: {
       bash: {
-        allow: ["npm *", "pnpm *", "npx *", "git *", "node *", "python *", "tsx *"]
+        "*": "ask",
+        "npm *": "allow",
+        "pnpm *": "allow",
+        "npx *": "allow",
+        "git *": "allow",
+        "node *": "allow",
+        "tsx *": "allow",
+        "ls *": "allow",
+        "mkdir *": "allow",
+        "rm *": "deny",
+        "rm -rf *": "deny",
+        "rm -r *": "deny",
+        "git push *-f*": "deny",
+        "git push *--force*": "deny"
       },
       edit: {
-        deny: [".opencode/agents/*"]
-      },
-      task: {
-        allow: agentAllowList
+        "*": "allow",
+        ".opencode/agents/*": "deny"
       },
-      external_directory: {
-        allow: ["/var/folders/**"]
+      read: {
+        "*": "allow",
+        "*.env": "deny",
+        "*.env.*": "deny",
+        "*.env.example": "allow"
+      }
+    },
+    mcp: {
+      context7: {
+        type: "remote",
+        url: "https://mcp.context7.com/mcp"
       }
     }
   };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ai-squad",
-  "version": "0.1.0",
+  "version": "0.2.1",
   "description": "Interactive CLI to scaffold AI agent infrastructure for new projects",
   "type": "module",
   "main": "./dist/index.js",
@@ -37,10 +37,10 @@
     "scaffold",
     "cli"
   ],
-  "author": "Pourya Fekri",
+  "author": "Pourya Akraminayeri",
   "repository": {
     "type": "git",
-    "url": "https://github.com/pouryafekri/ai-squad"
+    "url": "https://github.com/Pouryaak/ai-squad"
   },
   "engines": {
     "node": ">=18.0.0"

package/sections/shared/frontmatter/teamlead.yaml

@@ -5,4 +5,8 @@ description: >-
   Delegation-only — never writes code, never fixes issues directly.
 mode: primary
 hidden: false
-color: "#D4845F"
+color: "#D4845F"
+permission:
+  task:
+    "*": "deny"
+    "squad-*": "allow"