specweave 0.6.7 → 0.7.0

package/src/adapters/codex/adapter.ts

@@ -1,333 +0,0 @@
-/**
- * Codex Adapter (OpenAI)
- *
- * Semi-automation adapter for OpenAI Codex (ChatGPT Code Interpreter/Codex CLI).
- * Uses universal AGENTS.md file for instructions.
- *
- * Codex features:
- * - Works in CLI, IDE, web, GitHub, iOS app
- * - GPT-5-Codex optimized for engineering tasks
- * - File read/write operations
- * - Test execution and validation
- * - Task-based isolated environments
- */
-
-import * as path from 'path';
-import fs from 'fs-extra';
-import { AdapterBase } from '../adapter-base.js';
-import { AdapterOptions, AdapterFile } from '../adapter-interface.js';
-import { AgentsMdGenerator } from '../agents-md-generator.js';
-import { getDirname } from '../../utils/esm-helpers.js';
-
-const __dirname = getDirname(import.meta.url);
-
-export class CodexAdapter extends AdapterBase {
-  name = 'codex';
-  description = 'OpenAI Codex adapter - Semi-automation with AGENTS.md and task-based execution';
-  automationLevel = 'semi' as const;
-
-  /**
-   * Detect if Codex is available
-   *
-   * Checks for:
-   * - codex command in PATH (Codex CLI)
-   * - .codex/ directory exists
-   * - AGENTS.md file exists (previously configured)
-   */
-  async detect(): Promise<boolean> {
-    const hasCodexCLI = await this.commandExists('codex');
-    const hasCodexDir = await this.fileExists('.codex');
-    const hasAgentsMd = await this.fileExists('AGENTS.md');
-
-    return hasCodexCLI || hasCodexDir || hasAgentsMd;
-  }
-
-  /**
-   * Get files to install for Codex adapter
-   */
-  getFiles(): AdapterFile[] {
-    return [
-      {
-        sourcePath: 'AGENTS.md',
-        targetPath: 'AGENTS.md',
-        description: 'Universal SpecWeave instructions (works with all AI tools)'
-      },
-      {
-        sourcePath: 'README.md',
-        targetPath: '.codex/README.md',
-        description: 'Codex adapter documentation'
-      }
-    ];
-  }
-
-  /**
-   * Install Codex adapter
-   */
-  async install(options: AdapterOptions): Promise<void> {
-    console.log('\n📦 Installing OpenAI Codex Adapter (Semi-Automation)\n');
-
-    // Ensure .codex directory exists
-    const codexDir = path.join(options.projectPath, '.codex');
-    await fs.ensureDir(codexDir);
-
-    // Generate AGENTS.md
-    const agentsMdPath = path.join(options.projectPath, 'AGENTS.md');
-    await this.generateAgentsMd(agentsMdPath, options);
-
-    // Copy README
-    const readmePath = path.join(__dirname, 'README.md');
-    if (await fs.pathExists(readmePath)) {
-      await fs.copy(readmePath, path.join(codexDir, 'README.md'));
-    }
-
-    console.log('\n✨ Codex adapter installed!');
-    console.log('\n📋 Files created:');
-    console.log('   - AGENTS.md (universal instructions)');
-    console.log('   - .codex/README.md (adapter documentation)');
-  }
-
-  /**
-   * Generate AGENTS.md file
-   */
-  private async generateAgentsMd(targetPath: string, options: AdapterOptions): Promise<void> {
-    const generator = new AgentsMdGenerator(
-      path.join(__dirname, '../../skills'),
-      path.join(__dirname, '../../agents'),
-      path.join(__dirname, '../../commands')
-    );
-
-    const content = await generator.generate({
-      projectName: options.projectName,
-      projectPath: options.projectPath
-    });
-
-    await fs.writeFile(targetPath, content);
-    console.log('  ✅ AGENTS.md - Universal SpecWeave instructions');
-  }
-
-  /**
-   * Post-installation instructions
-   */
-  async postInstall(options: AdapterOptions): Promise<void> {
-    console.log(this.getInstructions());
-  }
-
-  /**
-   * Get usage instructions for Codex adapter
-   */
-  getInstructions(): string {
-    return `
-================================================================
-      OpenAI Codex Adapter - Semi-Automation
-================================================================
-
-Your project is now configured for OpenAI Codex with SEMI-automation!
-
-WHAT THIS PROVIDES:
-
-- AGENTS.md (Universal Instructions)
-  - Works with Codex CLI, ChatGPT web, and IDEs
-  - Auto-generated from actual skills/agents
-  - SpecWeave structure and workflows
-  - GPT-5-Codex optimized guidance
-
-- Task-Based Execution
-  - Isolated environments per task
-  - File read/write operations
-  - Test execution and validation
-
-QUICK START:
-
-**Option 1: Codex CLI** (Fastest)
-
-1. Install Codex CLI:
-
-   # Included with ChatGPT Plus, Pro, Business, Enterprise
-   npm install -g openai-codex-cli
-
-2. Create your first feature:
-
-   codex "Read AGENTS.md and create increment for user authentication"
-
-   Codex will:
-   - Read AGENTS.md for SpecWeave context
-   - Adopt PM role: Create spec.md (WHAT/WHY)
-   - Adopt Architect role: Create plan.md (HOW)
-   - Create tasks.md
-   - Execute in isolated environment
-
-**Option 2: ChatGPT Web** (Most Accessible)
-
-1. Open ChatGPT (Plus/Pro/Business/Enterprise)
-
-2. Upload AGENTS.md file
-
-3. Say:
-   "I'm using SpecWeave framework. Create increment for user authentication."
-
-4. Copy generated content to files:
-   - spec.md
-   - plan.md
-   - tasks.md
-
-**Option 3: IDE Integration** (If available in your IDE)
-
-Use Codex integration in your IDE (VS Code, JetBrains, etc.)
-Reference AGENTS.md for SpecWeave structure.
-
-HOW TO USE:
-
-**With Codex CLI**:
-\`\`\`bash
-# Explicit reference (recommended):
-codex "Read AGENTS.md and [your request]"
-
-# Natural language:
-codex "Create a new feature following SpecWeave structure"
-
-# With context:
-codex "Following the PM role in AGENTS.md, create requirements for payments"
-\`\`\`
-
-**With ChatGPT Web**:
-1. Upload AGENTS.md at start of conversation
-2. Reference it: "Following AGENTS.md, create increment for..."
-3. Download or copy-paste generated files
-
-CONTEXT LOADING (70%+ Token Savings):
-
-CRITICAL: Tell Codex to read context-manifest.yaml first!
-
-\`\`\`bash
-# CLI:
-codex "Read context-manifest.yaml from increment 0001, load only those files, then implement task T001"
-
-# Web:
-# 1. Upload context-manifest.yaml
-# 2. Upload only files listed in manifest
-# 3. Request task implementation
-\`\`\`
-
-FEATURES:
-
-✅ GPT-5-Codex Model
-   - Optimized for engineering tasks
-   - Trained on complex, real-world projects
-   - Code generation, debugging, refactoring
-
-✅ Task-Based Execution
-   - Each task in isolated environment
-   - Real-time progress monitoring
-   - Verifiable evidence (citations, logs)
-
-✅ File Operations
-   - Read and edit files
-   - Run commands (tests, linters, type checkers)
-   - Create new files and directories
-
-✅ Multiple Access Points
-   - CLI (fastest)
-   - Web (most accessible)
-   - IDE (most integrated)
-   - GitHub (for PRs)
-   - iOS app (mobile)
-
-✅ Test Execution
-   - Run test harnesses
-   - Validate implementations
-   - Ensure quality
-
-EXAMPLE WORKFLOWS:
-
-1. **Create Feature (CLI)**:
-   \`\`\`bash
-   codex "Read AGENTS.md. Create increment 0002 for payment processing with Stripe. Follow SpecWeave structure."
-   \`\`\`
-
-2. **Create Feature (Web)**:
-   - Upload AGENTS.md
-   - Say: "Create increment for payment processing"
-   - Download generated files
-
-3. **Implement Task**:
-   \`\`\`bash
-   codex "Read increment 0002, implement task T001 (Setup Stripe SDK)"
-   \`\`\`
-
-4. **Fix Bug**:
-   \`\`\`bash
-   codex "Read AGENTS.md and increment 0001. Fix authentication bug. Run tests."
-   \`\`\`
-
-5. **Code Review**:
-   \`\`\`bash
-   codex "Adopt Tech Lead role from AGENTS.md. Review src/auth/ code."
-   \`\`\`
-
-TIPS & TRICKS:
-
-1. **Always Reference AGENTS.md**
-   - CLI: Explicit in command
-   - Web: Upload at conversation start
-
-2. **Use Task-Based Approach**
-   - Break work into tasks (T001, T002, etc.)
-   - Each task runs in isolated environment
-   - Easier to track and validate
-
-3. **Monitor Progress**
-   - CLI shows real-time progress (1-30 minutes per task)
-   - Web shows thinking process
-   - Review citations and logs
-
-4. **Leverage Multiple Access Points**
-   - CLI for speed
-   - Web for accessibility
-   - IDE for integration
-
-5. **Validate with Tests**
-   - Codex can run tests automatically
-   - Verify implementations work
-
-COMPARISON:
-
-| Feature | Claude Code | Codex |
-|---------|-------------|-------|
-| **Automation** | Full | Semi |
-| **Skills** | Native | Via AGENTS.md |
-| **Agents** | Native | Via AGENTS.md |
-| **Access** | CLI only | CLI + Web + IDE + GitHub + iOS |
-| **Model** | Sonnet 4.5 | GPT-5-Codex |
-| **Task Isolation** | No | Yes (isolated environments) |
-
-LIMITATIONS:
-
-- No native skills/agents (manual role adoption via AGENTS.md)
-- No hooks (can't auto-update docs on events)
-- Tasks run 1-30 minutes (not instant)
-- Requires ChatGPT Plus/Pro/Business/Enterprise
-
-But very powerful with GPT-5-Codex and multiple access points!
-
-PLANS & PRICING:
-
-- **ChatGPT Plus**: $20/month (includes Codex CLI + Web)
-- **ChatGPT Pro**: $200/month (unlimited, faster)
-- **Business/Enterprise**: Custom pricing
-
-Free trial may be available - check OpenAI website.
-
-DOCUMENTATION:
-
-- AGENTS.md: Universal SpecWeave instructions (READ THIS!)
-- SPECWEAVE.md: Complete framework documentation
-- .codex/README.md: Codex adapter guide
-- https://openai.com/codex/
-
-You're ready to build with SpecWeave on OpenAI Codex!
-
-Pro tip: Use CLI for speed, Web for accessibility, and always
-reference AGENTS.md for SpecWeave structure!
-    `;
-  }
-}

package/src/adapters/copilot/README.md

@@ -1,240 +0,0 @@
-# GitHub Copilot Adapter
-
-**Automation Level**: Basic (Code suggestions with workspace context)
-
-## Overview
-
-The GitHub Copilot adapter provides **basic automation** for SpecWeave by configuring workspace instructions that Copilot reads to provide better code suggestions.
-
-## Key Limitation: No Custom Commands
-
-Unlike Claude Code (native skills/agents/commands) and Cursor (simulated via .cursorrules), GitHub Copilot does NOT support:
-- Custom slash commands
-- Workflow automation
-- Role-based assistance
-- Auto-activation
-
-**What it DOES support**:
-- Workspace instructions (.github/copilot/instructions.md)
-- Better code suggestions based on project context
-- Copilot Chat for Q&A
-- **Progressive disclosure via SKILLS-INDEX.md** (NEW in v0.3.8) - Access all 35+ skills!
-
-## What This Adapter Provides
-
-### 🔍 Progressive Disclosure (NEW)
-
-SpecWeave now includes **SKILLS-INDEX.md** - enabling GitHub Copilot to discover and use all 35+ skills!
-
-**How it works**:
-1. Read `.claude/skills/SKILLS-INDEX.md` (referenced in AGENTS.md)
-2. Match task to activation keywords
-3. Load specific SKILL.md when relevant
-4. Follow proven workflows
-
-**Benefits**:
-- ✅ **90% token savings** (5k vs 50k tokens)
-- ✅ **Full skill access** - All 35+ skills now available to Copilot!
-- ✅ **Consistent output** - Follow SpecWeave best practices
-
-**Example**:
-```markdown
-# From AGENTS.md, you learn about:
-.claude/skills/SKILLS-INDEX.md  # Single-file skill reference
-
-# When user asks: "Plan a new feature"
-# 1. Read SKILLS-INDEX.md → Find "increment-planner" skill
-# 2. Load .claude/skills/increment-planner/SKILL.md
-# 3. Follow the increment planning workflow
-```
-
-### .github/copilot/instructions.md
-- **What**: Workspace guidance that Copilot reads automatically
-- **Contains**: Project structure, workflows, best practices
-- **Benefit**: Copilot suggests code following SpecWeave patterns
-
-**Example**:
-```markdown
-# In instructions.md:
-## Workflow: Creating Features
-
-1. Create increment folder: .specweave/increments/####-feature-name/
-2. Create spec.md (WHAT & WHY - technology-agnostic)
-3. Create plan.md (HOW - technology-specific)
-4. Create tasks.md (implementation checklist)
-5. Create context-manifest.yaml (context loading - 70%+ token savings)
-```
-
-When you start typing in spec.md, Copilot suggests content following this structure!
-
-## Installation
-
-```bash
-# Install SpecWeave with Copilot adapter
-npx specweave init my-project --adapter copilot
-
-# Files created:
-# .github/copilot/instructions.md    (workspace instructions)
-```
-
-## Directory Structure
-
-```
-.github/copilot/
-├── instructions.md    # Workspace instructions (Copilot reads this)
-└── README.md          # This file
-```
-
-## Usage Examples
-
-### Example 1: Creating spec.md
-
-**You**:
-1. Create file: `.specweave/increments/0001-auth/spec.md`
-2. Start typing YAML frontmatter: `---`
-
-**Copilot** (based on instructions.md):
-```markdown
----
-increment: 0001-auth
-title: "User Authentication"
-priority: P1
-status: planned
----
-
-# Increment 0001: User Authentication
-
-## Overview
-...
-```
-
-Copilot suggests the complete structure!
-
-### Example 2: Copilot Chat Q&A
-
-**You**: "How do I create a plan.md following SpecWeave?"
-
-**Copilot** (reads instructions.md):
-```
-Based on SpecWeave guidelines, plan.md should contain:
-
-1. Architecture section (components, data model, APIs)
-2. Technology-specific details (unlike spec.md which is tech-agnostic)
-3. References to ADRs (Architecture Decision Records)
-4. Implementation strategy
-
-Example structure:
-...
-```
-
-### Example 3: Context Manifest
-
-**You**: Start typing in `context-manifest.yaml`
-
-**Copilot** (based on patterns in instructions.md):
-```yaml
----
-spec_sections:
-  - .specweave/docs/internal/strategy/auth/spec.md
-documentation:
-  - .specweave/docs/internal/architecture/auth/design.md
-max_context_tokens: 10000
----
-```
-
-## Comparison with Other Adapters
-
-| Feature | Claude Code | Cursor | Copilot | Generic |
-|---------|-------------|--------|---------|---------|
-| **Automation** | Full | Semi | Basic | Manual |
-| **Skills** | Native | Simulated | N/A | N/A |
-| **Agents** | Native | Manual roles | N/A | N/A |
-| **Commands** | Slash commands | Workflow instructions | N/A | Step-by-step guide |
-| **Context Shortcuts** | Native | @ shortcuts | N/A | Manual copy-paste |
-| **Code Suggestions** | ✅ | ✅ | ✅ | ❌ |
-| **Workflow Automation** | ✅ | ⚙️ | ❌ | ❌ |
-
-**Copilot = Basic but useful** (better than nothing!)
-
-## Limitations
-
-❌ **No workflow automation** - Must create folders/files manually
-❌ **No skills/agents** - Can't simulate roles like Cursor
-❌ **No commands** - No slash commands or @ shortcuts
-❌ **No hooks** - Can't auto-update docs (see workaround below)
-⚠️ **Completely manual workflow** - Just better code suggestions
-
-✅ **But still helpful for**:
-- Code completion following SpecWeave patterns
-- Suggesting file structures
-- Copilot Chat Q&A about project
-
-### Documentation Update Workaround
-
-Since GitHub Copilot doesn't have hooks, you MUST manually update documentation after every task.
-
-**See the comprehensive guide in AGENTS.md** (section: "Documentation Updates - CRITICAL FOR NON-CLAUDE TOOLS")
-
-**Quick checklist after completing any task**:
-1. Update `.specweave/increments/{id}/tasks.md` (mark tasks complete)
-2. Update `.specweave/docs/internal/architecture/` (HLD/LLD/ADRs)
-3. Update `.specweave/docs/internal/strategy/` (PRDs if requirements changed)
-4. Update `README.md` (user-facing changes)
-5. Update `CHANGELOG.md` (version history)
-
-## When to Use This Adapter
-
-✅ **Use Copilot adapter if**:
-- You already use VS Code + GitHub Copilot
-- You want better code suggestions
-- You don't mind manual workflows
-- Simple projects
-
-⚠️ **Consider alternatives if**:
-- You want automation → Use Claude Code (full) or Cursor (semi)
-- You need skills/agents → Use Claude Code
-- You want role-based assistance → Use Claude Code or Cursor
-
-## How Copilot Uses Workspace Instructions
-
-**Automatic (no action needed)**:
-1. You open project in VS Code with Copilot
-2. Copilot reads `.github/copilot/instructions.md` automatically
-3. Copilot understands:
-   - SpecWeave structure (.specweave/ folders)
-   - File naming (spec.md, plan.md, tasks.md)
-   - Patterns (context manifests, test IDs)
-   - Best practices (tech-agnostic specs, etc.)
-
-**Result**: Better code suggestions that fit SpecWeave!
-
-**Example**:
-- Type `# Increment` → Copilot suggests increment structure
-- Type `TC-0001` → Copilot suggests test case format
-- Type `context-manifest.yaml` → Copilot suggests manifest structure
-
-## Tips & Tricks
-
-### 1. Reference instructions.md
-If Copilot suggests wrong patterns: "Follow .github/copilot/instructions.md"
-
-### 2. Use Copilot Chat
-Ask: "How does SpecWeave work?" → Copilot explains from instructions
-
-### 3. Start Typing, Let Copilot Complete
-Begin with YAML frontmatter or headers → Copilot fills in rest
-
-### 4. Check Context Manifests
-Open context-manifest.yaml → Copilot sees relevant files
-
-## Related Documentation
-
-- [SPECWEAVE.md](../../SPECWEAVE.md) - Complete development guide
-- [.github/copilot/instructions.md](.github/copilot/instructions.md) - Workspace instructions
-- [Adapter Architecture](../README.md) - Multi-tool design philosophy
-
----
-
-**Status**: Active (v0.2.0-beta.1+)
-**Market Share**: ~40% (GitHub Copilot users)
-**Priority**: P1 (large market, even with basic automation)