ai-sdlc 0.1.0-alpha.1 → 0.1.0-alpha.11

package/README.md

@@ -1,4 +1,4 @@
-# Agentic SDLC
+# AI SDLC
 
 > **Alpha Release**: This is an early alpha. Expect breaking changes.
 > Report issues at [GitHub Issues](https://github.com/patrob/agentic-workflow/issues)
@@ -7,14 +7,15 @@ Agent-first SDLC workflow manager using Claude Agent SDK. A Kanban-style board w
 
 ## Features
 
-- 📋 Kanban-style story management (Backlog → Ready → In Progress → Done)
-- 🤖 AI-powered agents for each workflow stage
-- 🚀 **Full SDLC automation** for individual stories with `--auto --story` (refine → research → plan → implement → review)
-- 🔄 **Resume workflows after interruption** with `--continue` flag
-- 🎨 Customizable themes (auto, light, dark, none)
-- 📊 Visual progress tracking with status flags
-- ⚡ Automatic state assessment and action recommendations
-- ♻️ Smart phase skipping (automatically skips completed phases)
+- 📋 **Kanban-style story management** (Backlog → Ready → In Progress → Done)
+- 🤖 **AI-powered agents** for each workflow stage (refine, research, plan, implement, review)
+- 🚀 **Full SDLC automation** with `--auto --story` - takes a story from idea to reviewed code
+- 🔴🟢🔵 **TDD Mode** - Optional Test-Driven Development enforcement with Red-Green-Refactor cycles
+- 🔄 **Resume workflows** after interruption with `--continue` flag
+- 👀 **Daemon mode** - Continuously watch for and process new stories with `--watch`
+- 🎨 **Customizable themes** (auto, light, dark, none)
+- 📊 **Visual progress tracking** with status flags [R][P][I][V]
+- ♻️ **Smart phase skipping** - automatically skips completed phases
 
 ## Installation
 
@@ -26,36 +27,47 @@ npm run build
 ## Quick Start
 
 ```bash
-# Initialize the .agentic-sdlc folder
-agentic-sdlc init
+# 1. Initialize the project
+ai-sdlc init
 
-# Add a new story
-agentic-sdlc add "Implement user authentication"
-
-# Run the workflow (process next action)
-agentic-sdlc run
+# 2. Add a story to the backlog
+ai-sdlc add "Implement user authentication"
 
-# Process all pending actions automatically
-agentic-sdlc run --auto
+# 3. View your board
+ai-sdlc status
 
-# Resume workflow after interruption
-agentic-sdlc run --continue
+# 4. Run the full SDLC for a story (refine → research → plan → implement → review)
+ai-sdlc run --auto --story implement-user-authentication
 ```
 
+### Common Commands
+
+| Command | Description |
+|---------|-------------|
+| `ai-sdlc status` | View all stories in Kanban board |
+| `ai-sdlc add "title"` | Add a new story to backlog |
+| `ai-sdlc run` | Process next recommended action |
+| `ai-sdlc run --auto` | Process all pending actions |
+| `ai-sdlc run --auto --story <id>` | Full SDLC for one story |
+| `ai-sdlc run --continue` | Resume after interruption |
+| `ai-sdlc run --watch` | Daemon mode - watch for new stories |
+| `ai-sdlc details <id>` | Show story details |
+| `ai-sdlc config` | View/set configuration |
+
 ## CLI Commands
 
 ### `init`
-Initialize the `.agentic-sdlc` folder structure with Kanban columns.
+Initialize the `.ai-sdlc` folder structure with Kanban columns.
 
 ```bash
-agentic-sdlc init
+ai-sdlc init
 ```
 
 ### `status`
 View all stories in a formatted table view with story IDs, truncated text, and uniform alignment.
 
 ```bash
-agentic-sdlc status
+ai-sdlc status
 ```
 
 **Table View** (terminal width ≥ 100 columns):
@@ -91,13 +103,13 @@ Labels: enhancement, ui, cli | Flags: [R]
 
 **Minimum Terminal Width**: 100 columns recommended for table view
 
-**Disable Hints**: Set `AGENTIC_SDLC_NO_HINTS=1` to hide the compact view notification
+**Disable Hints**: Set `AI_SDLC_NO_HINTS=1` to hide the compact view notification
 
 ### `add <title>`
 Add a new story to the backlog.
 
 ```bash
-agentic-sdlc add "Add dark mode toggle"
+ai-sdlc add "Add dark mode toggle"
 ```
 
 ### `details <id>` (alias: `d`)
@@ -107,13 +119,13 @@ Show detailed information about a specific story by ID or slug.
 
 ```bash
 # View by story ID
-agentic-sdlc details story-mk68fjh7-fvbt
+ai-sdlc details story-mk68fjh7-fvbt
 
 # View by slug
-agentic-sdlc details add-dark-mode-toggle
+ai-sdlc details add-dark-mode-toggle
 
 # Use short alias
-agentic-sdlc d story-mk68fjh7-fvbt
+ai-sdlc d story-mk68fjh7-fvbt
 ```
 
 **Displays:**
@@ -139,27 +151,31 @@ Run the workflow and process actions.
 - `--continue` - Resume workflow from last checkpoint
 - `--story <id-or-slug>` - Target a specific story by ID or slug
 - `--step <phase>` - Run a specific phase (refine, research, plan, implement, review) - cannot be combined with `--auto --story`
+- `--watch` - Run in daemon mode, continuously processing backlog (NEW!)
 
 **Examples:**
 
 ```bash
 # Process the next recommended action
-agentic-sdlc run
+ai-sdlc run
 
 # Process all pending actions
-agentic-sdlc run --auto
+ai-sdlc run --auto
 
 # Resume after interruption (Ctrl+C, error, etc.)
-agentic-sdlc run --continue
+ai-sdlc run --continue
 
 # Preview what would be executed
-agentic-sdlc run --dry-run
+ai-sdlc run --dry-run
 
 # Run full SDLC for a specific story (NEW!)
-agentic-sdlc run --auto --story my-feature
+ai-sdlc run --auto --story my-feature
 
 # Run specific phase for a story
-agentic-sdlc run --story my-feature --step research
+ai-sdlc run --story my-feature --step research
+
+# Run in daemon/watch mode (NEW!)
+ai-sdlc run --watch
 ```
 
 ### `config [key] [value]`
@@ -167,13 +183,13 @@ Manage configuration settings.
 
 ```bash
 # View all configuration
-agentic-sdlc config
+ai-sdlc config
 
 # View theme setting
-agentic-sdlc config theme
+ai-sdlc config theme
 
 # Set theme
-agentic-sdlc config theme dark
+ai-sdlc config theme dark
 ```
 
 ## Full SDLC Automation (--auto --story)
@@ -184,7 +200,7 @@ The `--auto --story` combination provides complete end-to-end automation for ind
 
 ```bash
 # Take a story from idea to reviewed implementation in one command
-agentic-sdlc run --auto --story my-feature
+ai-sdlc run --auto --story my-feature
 ```
 
 ### How It Works
@@ -225,7 +241,7 @@ Already completed phases are automatically skipped:
 
 ```bash
 # Story has research and plan complete
-agentic-sdlc run --auto --story my-feature
+ai-sdlc run --auto --story my-feature
 
 # Output:
 # 🚀 Starting full SDLC for story: My Feature
@@ -292,7 +308,7 @@ Fix the error above and use --continue to resume.
 
 **Recovery:**
 1. Fix the underlying issue
-2. Resume with: `agentic-sdlc run --continue`
+2. Resume with: `ai-sdlc run --continue`
 3. Workflow continues from failed phase
 
 ### Resuming Full SDLC Workflows
@@ -301,12 +317,12 @@ Full SDLC mode integrates seamlessly with checkpoint/resume:
 
 ```bash
 # Start full SDLC workflow
-agentic-sdlc run --auto --story my-feature
+ai-sdlc run --auto --story my-feature
 
 # (Interrupted during implementation phase)
 
 # Resume automatically - full SDLC mode is restored
-agentic-sdlc run --continue
+ai-sdlc run --continue
 
 # Output:
 # ⟳ Resuming workflow from checkpoint
@@ -328,13 +344,13 @@ Full SDLC mode respects configured stage gates:
 
 ```bash
 # With requireApprovalBeforeImplementation enabled
-agentic-sdlc run --auto --story my-feature
+ai-sdlc run --auto --story my-feature
 
 # Output:
 # ✓ Research complete
 # ✓ Plan complete
 # ⚠️ Stage gate: Implementation requires approval
-# Run 'agentic-sdlc run --continue' to proceed after approval
+# Run 'ai-sdlc run --continue' to proceed after approval
 ```
 
 ### Review Retry Logic
@@ -359,10 +375,10 @@ The system tracks retry attempts and stops if max retries are exceeded.
 
 ```bash
 # Add a new story
-agentic-sdlc add "Add user authentication"
+ai-sdlc add "Add user authentication"
 
 # Run complete SDLC automation
-agentic-sdlc run --auto --story add-user-authentication
+ai-sdlc run --auto --story add-user-authentication
 
 # Story progresses through all phases automatically:
 # 1. Refine (backlog → ready)
@@ -386,12 +402,12 @@ agentic-sdlc run --auto --story add-user-authentication
 **Cannot combine with --step:**
 ```bash
 # This will error:
-agentic-sdlc run --auto --story my-feature --step research
+ai-sdlc run --auto --story my-feature --step research
 
 # Error: Cannot combine --auto --story with --step flag.
 # Use either:
-#   - agentic-sdlc run --auto --story my-feature (full SDLC)
-#   - agentic-sdlc run --story my-feature --step research (single phase)
+#   - ai-sdlc run --auto --story my-feature (full SDLC)
+#   - ai-sdlc run --story my-feature --step research (single phase)
 ```
 
 **All phases are executed in sequence:**
@@ -411,6 +427,267 @@ agentic-sdlc run --auto --story my-feature --step research
 - ❌ Running just one specific phase (use `--story --step` instead)
 - ❌ Interactive workflows requiring manual review between phases
 
+## TDD Mode (Test-Driven Development)
+
+TDD mode enforces strict Test-Driven Development practices during implementation, ensuring code is developed using the Red-Green-Refactor cycle. This helps eliminate review failures caused by missing or improperly sequenced tests.
+
+### Enabling TDD Mode
+
+**Per-story** (recommended for gradual adoption):
+
+Add `tdd_enabled: true` to your story's frontmatter:
+
+```yaml
+---
+id: story-abc123
+title: Add user authentication
+status: ready
+tdd_enabled: true
+---
+```
+
+**Globally** (for all stories):
+
+Add to your `.ai-sdlc.json` config:
+
+```json
+{
+  "tdd": {
+    "enabled": true
+  }
+}
+```
+
+### How TDD Mode Works
+
+When TDD is enabled, the implementation agent follows a strict Red-Green-Refactor cycle for each acceptance criterion:
+
+```
+┌─────────────────────────────────────────────────────────────────┐
+│                    TDD Cycle (per AC)                           │
+├─────────────────────────────────────────────────────────────────┤
+│  🔴 RED      → Write failing test    → Verify test FAILS        │
+│  🟢 GREEN    → Write minimum code    → Verify test PASSES       │
+│              →                       → Verify NO regressions    │
+│  🔵 REFACTOR → Improve code quality  → Verify ALL tests pass    │
+└─────────────────────────────────────────────────────────────────┘
+```
+
+**Cycle enforcement:**
+1. **RED**: Agent writes a test for the next acceptance criterion, verifies it fails
+2. **GREEN**: Agent writes minimum code to pass, verifies test passes and no regressions
+3. **REFACTOR**: Agent improves code quality, verifies all tests still pass
+4. **Record**: Cycle is recorded to story frontmatter and persisted
+5. **Repeat**: Process continues until all acceptance criteria are covered
+
+### TDD Validation in Review
+
+When TDD is enabled, the review agent validates that:
+- All TDD cycles have complete RED → GREEN → REFACTOR phases
+- All tests remained green after each phase
+- No cycles were skipped or incomplete
+
+Violations generate critical review issues that must be addressed.
+
+### TDD Configuration Options
+
+```json
+{
+  "tdd": {
+    "enabled": false,
+    "strictMode": true,
+    "maxCycles": 50,
+    "requireApprovalPerCycle": false
+  }
+}
+```
+
+| Option | Default | Description |
+|--------|---------|-------------|
+| `enabled` | `false` | Enable TDD mode globally (can be overridden per-story) |
+| `strictMode` | `true` | Reserved for future strict enforcement rules |
+| `maxCycles` | `50` | Maximum TDD cycles before stopping (prevents infinite loops) |
+| `requireApprovalPerCycle` | `false` | Reserved for future approval workflow |
+
+### TDD Cycle History
+
+TDD cycles are recorded in the story frontmatter for audit and debugging:
+
+```yaml
+tdd_test_history:
+  - test_name: "should authenticate valid user"
+    test_file: "src/auth/login.test.ts"
+    red_timestamp: "2024-01-15T10:00:00.000Z"
+    green_timestamp: "2024-01-15T10:05:00.000Z"
+    refactor_timestamp: "2024-01-15T10:08:00.000Z"
+    all_tests_green: true
+    cycle_number: 1
+```
+
+History is trimmed to the last 100 cycles to prevent unbounded growth.
+
+### When to Use TDD Mode
+
+**Ideal for:**
+- ✅ New feature development where test coverage is critical
+- ✅ Teams adopting TDD practices
+- ✅ Stories with clear, testable acceptance criteria
+- ✅ Reducing review failures from missing tests
+
+**Consider standard mode for:**
+- ❌ Quick fixes or hotfixes where speed is critical
+- ❌ Refactoring tasks without new functionality
+- ❌ Stories without clear testable criteria
+
+## Daemon/Watch Mode (--watch)
+
+The `--watch` flag runs AI-SDLC in daemon mode, continuously monitoring the backlog folder for new stories and automatically processing them through the full workflow pipeline.
+
+### Quick Example
+
+```bash
+# Start daemon mode - runs indefinitely
+ai-sdlc run --watch
+```
+
+### How It Works
+
+When you run with `--watch`, the system:
+
+1. **Starts a file system watcher** monitoring `.ai-sdlc/backlog/*.md`
+2. **Detects new story files** in real-time using chokidar
+3. **Automatically processes each story** through the workflow (refine → research → plan → implement → review)
+4. **Queues multiple stories** for sequential processing (no parallel execution)
+5. **Continues running** even if individual stories fail
+6. **Logs all activity** to console for monitoring
+
+### Usage
+
+```bash
+# Start daemon mode
+ai-sdlc run --watch
+
+# Output:
+# 🤖 AI-SDLC Daemon Mode Started
+# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+#    SDLC Root: /path/to/.ai-sdlc
+#    Watching: .ai-sdlc/backlog/*.md
+#    Polling Interval: 5000ms
+# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+#
+# 👀 Watching for new stories...
+#    Press Ctrl+C to shutdown gracefully
+#
+# 📄 New story detected: my-feature.md
+#    ▶️  Starting workflow for: my-feature
+#    ✅ Workflow completed: my-feature
+#
+# 👀 Queue empty, waiting for new stories...
+```
+
+### Graceful Shutdown
+
+The daemon supports graceful shutdown to ensure stories complete processing:
+
+```bash
+# Single Ctrl+C - graceful shutdown
+Press Ctrl+C
+
+# Output:
+# 🛑 Shutting down gracefully...
+#    Waiting for current story to complete...
+#    Current story completed
+#    File watcher stopped
+# ✨ Daemon shutdown complete
+
+# Double Ctrl+C (within 2 seconds) - force quit
+Press Ctrl+C (twice quickly)
+
+# Output:
+# ⚠️  Force quitting...
+```
+
+**Shutdown Behavior:**
+- **First Ctrl+C**: Initiates graceful shutdown, waits for current story to complete (max 30 seconds)
+- **Second Ctrl+C**: Forces immediate exit if pressed within 2 seconds
+- **SIGTERM**: Also triggers graceful shutdown (useful for systemd/docker)
+
+### Configuration
+
+Daemon behavior can be customized in `.ai-sdlc.json`:
+
+```json
+{
+  "daemon": {
+    "enabled": false,
+    "pollingInterval": 5000,
+    "watchPatterns": [".ai-sdlc/backlog/*.md"],
+    "processDelay": 500,
+    "shutdownTimeout": 30000,
+    "enableEscShutdown": false,
+    "escTimeout": 500
+  }
+}
+```
+
+**Configuration Options:**
+- `enabled` - Enable daemon mode by default (default: `false`)
+- `pollingInterval` - Fallback polling interval in milliseconds (default: `5000`)
+- `watchPatterns` - Glob patterns to watch for new stories (default: `[".ai-sdlc/backlog/*.md"]`)
+- `processDelay` - Debounce delay for file changes in milliseconds (default: `500`)
+- `shutdownTimeout` - Maximum time to wait for graceful shutdown in milliseconds (default: `30000`)
+- `enableEscShutdown` - Enable Esc+Esc shutdown (Phase 2 feature, default: `false`)
+- `escTimeout` - Maximum time between Esc presses in milliseconds (default: `500`)
+
+### Error Handling
+
+The daemon is designed to be resilient:
+
+- **Story processing failures** are logged but don't stop the daemon
+- **Malformed story files** are skipped with error logging
+- **API failures** are logged and the story is marked with an error
+- **File system errors** are logged but the watcher continues
+- **Duplicate processing** is prevented via story ID tracking
+
+```bash
+# Example error handling
+📄 New story detected: malformed-story.md
+   ▶️  Starting workflow for: malformed-story
+   ❌ Workflow failed: malformed-story
+
+⚠️  Error processing malformed-story.md
+   Invalid YAML frontmatter
+   Daemon continues running...
+
+👀 Queue empty, waiting for new stories...
+```
+
+### Use Cases
+
+**Ideal for:**
+- ✅ Continuous integration environments
+- ✅ Monitoring backlog folders for new work
+- ✅ Unattended story processing
+- ✅ Development workflows with frequent story additions
+- ✅ Demo and testing scenarios
+
+**Not ideal for:**
+- ❌ Interactive development requiring manual approvals
+- ❌ Stories requiring mid-workflow intervention
+- ❌ Environments where resource usage must be minimal
+
+### Limitations (MVP)
+
+The current MVP implementation has these limitations:
+
+- **No Esc+Esc shutdown** - Only Ctrl+C is supported for shutdown
+- **No web dashboard** - All monitoring is via console output
+- **No multi-daemon support** - Only one daemon instance per SDLC folder
+- **No auto-restart** - Daemon must be restarted manually if it crashes
+- **Sequential processing only** - Stories are processed one at a time
+
+These features are planned for future releases.
+
 ## Resuming Workflows
 
 The `--continue` flag enables resuming workflows after interruption. This is useful when:
@@ -422,7 +699,7 @@ The `--continue` flag enables resuming workflows after interruption. This is use
 
 ### How It Works
 
-1. **Automatic Checkpointing**: After each successful action, the workflow state is saved to `.agentic-sdlc/.workflow-state.json`
+1. **Automatic Checkpointing**: After each successful action, the workflow state is saved to `.ai-sdlc/.workflow-state.json`
 2. **Smart Resume**: When you use `--continue`, the system:
    - Loads the saved checkpoint
    - Shows which actions were already completed
@@ -436,12 +713,12 @@ The `--continue` flag enables resuming workflows after interruption. This is use
 
 ```bash
 # Start workflow
-agentic-sdlc run --auto
+ai-sdlc run --auto
 
 # (Interrupted by Ctrl+C after research completes)
 
 # Resume from checkpoint
-agentic-sdlc run --continue
+ai-sdlc run --continue
 # Output:
 # ⟳ Resuming workflow from checkpoint
 #   Workflow ID: workflow-1234567890-abc123
@@ -459,14 +736,14 @@ agentic-sdlc run --continue
 
 ```bash
 # Resume and complete all remaining actions
-agentic-sdlc run --auto --continue
+ai-sdlc run --auto --continue
 ```
 
 **Check for existing checkpoint:**
 
 ```bash
 # Start a new workflow when checkpoint exists
-agentic-sdlc run
+ai-sdlc run
 # Output:
 # Note: Found previous checkpoint. Use --continue to resume.
 ```
@@ -505,7 +782,7 @@ Checkpoint cleared.
 
 Checkpoint files are stored at:
 ```
-.agentic-sdlc/.workflow-state.json
+.ai-sdlc/.workflow-state.json
 ```
 
 **State file format:**
@@ -519,12 +796,12 @@ Checkpoint files are stored at:
     {
       "type": "research",
       "storyId": "story-123",
-      "storyPath": ".agentic-sdlc/in-progress/add-dark-mode.md",
+      "storyPath": ".ai-sdlc/in-progress/add-dark-mode.md",
       "completedAt": "2024-01-08T15:45:12.000Z"
     }
   ],
   "context": {
-    "sdlcRoot": ".agentic-sdlc",
+    "sdlcRoot": ".ai-sdlc",
     "options": {
       "auto": true
     },
@@ -549,7 +826,7 @@ Each story progresses through these stages:
 
 ## Story Status Flags
 
-When viewing stories with `agentic-sdlc status`, you'll see progress flags:
+When viewing stories with `ai-sdlc status`, you'll see progress flags:
 
 - `R` - Research complete
 - `P` - Plan complete
@@ -565,30 +842,50 @@ IN-PROGRESS (1)
 
 ## Configuration
 
-Configuration is stored in `.agentic-sdlc.json` at the project root.
+Configuration is stored in `.ai-sdlc.json` at the project root.
 
 **Default configuration:**
 ```json
 {
-  "sdlcFolder": ".agentic-sdlc",
+  "sdlcFolder": ".ai-sdlc",
   "stageGates": {
     "requireApprovalBeforeImplementation": false,
     "requireApprovalBeforePR": false,
     "autoMergeOnApproval": false
   },
+  "tdd": {
+    "enabled": false,
+    "strictMode": true,
+    "maxCycles": 50,
+    "requireApprovalPerCycle": false
+  },
   "defaultLabels": [],
   "theme": "auto",
   "settingSources": []
 }
 ```
 
+**Configuration options:**
+
+| Section | Option | Default | Description |
+|---------|--------|---------|-------------|
+| `sdlcFolder` | - | `.ai-sdlc` | Folder for story files |
+| `stageGates` | `requireApprovalBeforeImplementation` | `false` | Pause before implementation |
+| `stageGates` | `requireApprovalBeforePR` | `false` | Pause before PR creation |
+| `stageGates` | `autoMergeOnApproval` | `false` | Auto-merge approved PRs |
+| `tdd` | `enabled` | `false` | Enable TDD mode globally |
+| `tdd` | `strictMode` | `true` | Reserved for future use |
+| `tdd` | `maxCycles` | `50` | Max TDD cycles per story |
+| `tdd` | `requireApprovalPerCycle` | `false` | Reserved for future use |
+| `theme` | - | `auto` | Color theme (auto/light/dark/none) |
+
 ### Project Settings with CLAUDE.md
 
 The Agent SDK can automatically load custom instructions from a `CLAUDE.md` file in your project's `.claude/` directory. This feature allows teams to share consistent AI behavior across all team members without manually specifying custom instructions.
 
 #### How to Enable
 
-Add `"project"` to the `settingSources` array in your `.agentic-sdlc.json`:
+Add `"project"` to the `settingSources` array in your `.ai-sdlc.json`:
 
 ```json
 {
@@ -613,7 +910,7 @@ your-project/
 │   ├── CLAUDE.md          # Custom instructions (you create this)
 │   ├── settings.json      # Project settings (SDK managed)
 │   └── settings.local.json # Local overrides (SDK managed, gitignored)
-├── .agentic-sdlc.json
+├── .ai-sdlc.json
 └── ...
 ```
 
@@ -753,7 +1050,7 @@ Available themes:
 
 ```bash
 # Set theme
-agentic-sdlc config theme dark
+ai-sdlc config theme dark
 ```
 
 ## Authentication
@@ -773,7 +1070,7 @@ export ANTHROPIC_API_KEY=your-key-here
 ## Folder Structure
 
 ```
-.agentic-sdlc/
+.ai-sdlc/
 ├── backlog/           # New stories
 ├── ready/             # Stories ready to start
 ├── in-progress/       # Active stories
@@ -822,14 +1119,14 @@ npm test -- --watch
 If you encounter a corrupted checkpoint error:
 
 ```bash
-Error: Corrupted workflow state file at .agentic-sdlc/.workflow-state.json.
-Delete the file to start fresh: rm ".agentic-sdlc/.workflow-state.json"
+Error: Corrupted workflow state file at .ai-sdlc/.workflow-state.json.
+Delete the file to start fresh: rm ".ai-sdlc/.workflow-state.json"
 ```
 
 Solution:
 ```bash
-rm .agentic-sdlc/.workflow-state.json
-agentic-sdlc run  # Start fresh
+rm .ai-sdlc/.workflow-state.json
+ai-sdlc run  # Start fresh
 ```
 
 ### Multiple Workflows
@@ -838,8 +1135,8 @@ Currently, only one workflow can run at a time. If you try to start a new workfl
 
 To start fresh (ignoring the checkpoint):
 ```bash
-rm .agentic-sdlc/.workflow-state.json
-agentic-sdlc run
+rm .ai-sdlc/.workflow-state.json
+ai-sdlc run
 ```
 
 ## License