ai-sprint-kit 2.1.29 → 2.1.31

package/README.md

@@ -6,7 +6,7 @@
 
 ## What is AI Sprint?
 
-AI Sprint transforms Claude Code into a team of 12 specialized AI agents with 15 powerful commands. Build production-grade software with security-first, autonomous development.
+AI Sprint transforms Claude Code into a team of 12 specialized AI agents with 17 powerful commands. Build production-grade software with security-first, autonomous development.
 
 ## What's New in v2.1
 
@@ -19,7 +19,7 @@ AI Sprint transforms Claude Code into a team of 12 specialized AI agents with 15
 ## Features
 
 - **12 AI Agents** - planner, implementer, tester, reviewer, security, devops, docs, debugger, researcher, verifier, browser-tester, deployment-smoke
-- **15 Commands** - /ai-sprint-plan, /ai-sprint-code, /ai-sprint-test, /ai-sprint-review, /ai-sprint-secure, /ai-sprint-deploy, /ai-sprint-docs, /ai-sprint-debug, /ai-sprint-scan, /ai-sprint-validate, /ai-sprint-browser-test, /ai-sprint-orchestrate, /ai-sprint-setup, /ai-sprint-auto
+- **17 Commands** - /ai-sprint-init, /ai-sprint-plan, /ai-sprint-code, /ai-sprint-test, /ai-sprint-review, /ai-sprint-secure, /ai-sprint-deploy, /ai-sprint-docs, /ai-sprint-debug, /ai-sprint-fix, /ai-sprint-scan, /ai-sprint-validate, /ai-sprint-browser-test, /ai-sprint-orchestrate, /ai-sprint-setup, /ai-sprint-auto, /ai-sprint-ralph
 - **Automation Hooks** - PostToolUse (auto-format, lint, sound), Stop (final verification), SubagentStart (context injection)
 - **Memory System** - Persistent context across sessions
 - **Security-First** - Built-in SAST, secrets scanning, dependency audit
@@ -86,6 +86,7 @@ Visit [ai-sprint-kit.app.data-espresso.com](https://ai-sprint-kit.app.data-espre
 ## Commands
 
 ```
+/ai-sprint-init          - Initialize project context (scan + CLAUDE.md)
 /ai-sprint-plan          - Create implementation plans
 /ai-sprint-code          - Generate or refactor code
 /ai-sprint-test          - Generate and run tests
@@ -94,14 +95,87 @@ Visit [ai-sprint-kit.app.data-espresso.com](https://ai-sprint-kit.app.data-espre
 /ai-sprint-deploy        - CI/CD setup and deployment
 /ai-sprint-docs          - Generate documentation
 /ai-sprint-debug         - Investigate issues
+/ai-sprint-fix           - Autonomous bug fix cycle
 /ai-sprint-scan          - Index codebase for AI context
 /ai-sprint-validate      - Pre-commit validation
 /ai-sprint-browser-test  - Browser automation tests (Playwright)
 /ai-sprint-orchestrate   - Multi-session coordination
 /ai-sprint-setup         - Configuration wizard
 /ai-sprint-auto          - Full autonomous development cycle
+/ai-sprint-ralph         - Persistent loop until task complete
 ```
 
+## Ralph Loop - Autonomous Task Completion
+
+Ralph runs tasks in a persistent loop until completion or max iterations reached. Ideal for migrations, batch fixes, and tasks with clear success criteria.
+
+### Simple Mode (Single Task)
+
+```bash
+# Fix all TypeScript errors
+/ai-sprint-ralph "fix all TypeScript errors" --promise "tsc: 0 errors"
+
+# Migrate tests to vitest
+/ai-sprint-ralph "migrate jest to vitest" --max-iterations 30
+
+# Add dark mode
+/ai-sprint-ralph "add dark mode toggle" --promise "DARK MODE DONE" --max-iterations 10
+```
+
+### PRD Mode (Multi-Feature)
+
+Create `ai_context/ralph/prd.md`:
+
+```markdown
+# Project: User Authentication
+
+## Features
+
+### 1. Login Form
+- [ ] Add login form component
+- [ ] Implement JWT validation
+- [ ] Add logout functionality
+
+### 2. API Endpoints
+- [ ] Create /api/users endpoint
+- [ ] Add authentication middleware
+```
+
+Then run:
+
+```bash
+/ai-sprint-ralph --prd ai_context/ralph/prd.md --max-iterations 50
+```
+
+Ralph works through each feature, checking off tasks as completed.
+
+### Options
+
+| Option | Default | Description |
+|--------|---------|-------------|
+| `--max-iterations N` | 20 | Maximum loop iterations |
+| `--promise STRING` | `<promise>COMPLETED</promise>` | Completion detection string |
+| `--prd FILE` | none | PRD file for multi-feature mode |
+| `--no-auto-commit` | false | Disable git commit after iteration |
+| `--resume` | false | Resume interrupted session |
+
+### Writing Good Prompts
+
+**Good prompts converge** - each iteration brings you closer to success:
+
+```
+Migrate all test files from Jest to Vitest.
+
+SUCCESS CRITERIA:
+1. All imports use "vitest" not "jest"
+2. npm run test exits with code 0
+3. Zero jest dependencies remain
+
+When ALL criteria met, output: <promise>MIGRATION COMPLETE</promise>
+```
+
+**Avoid vague goals** like "improve quality" - define measurable completion conditions.
+
 ## Support
 
 - GitHub Issues: [ai-sprint-pro/issues](https://github.com/apiasak/ai-sprint-pro/issues)

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ai-sprint-kit",
-  "version": "2.1.29",
+  "version": "2.1.31",
   "description": "CLI installer for AI Sprint autonomous development framework - requires license",
   "main": "lib/installer.js",
   "bin": {