Evidence-first incident diagnosis. Root cause in 16-30 seconds.
Paste an incident. Get root cause, fix plan, rollback steps, and tests. Backed by evidence, not guesses.
| Area | Progress | Detail |
|---|---|---|
| Overall | 100% | [████████████████████] |
| Core (4-agent, API, webhooks) | 100% | ✅ Shipped in v1.0.1 |
| Premium UI checkpoints | 29/29 (100%) | [████████████████████] |
What's done: Core pipeline, API, webhooks, audit trail, orchestration UI. Design tokens (colors, typography, motion). ThemeProvider, UIStateProvider, ThemeContext. AnimatedSection, FadeIn, motion-utils (prefersReducedMotion). Layout, ErrorBoundary, App.jsx. Dark/light themes. Skeleton, StepProgressBar wired in OrchestrationDashboard and LoadingOverlay.
What's left: None — all checkpoints complete.
Roadmap Features (from .github/PROJECT_1.0.0_CHECKPOINTS.md)
| Phase | Milestone | Scope |
|---|---|---|
| 1 | v1.1.0 Design Tokens & Motion | DT-001…DT-010: Colors, typography, motion, spacing, shadows, radius |
| 2 | v1.2.0 React Core & Migration | RC, RI, CM: App, ThemeProvider, UIStateProvider, Layout, ErrorBoundary |
| 3 | v1.3.0 Motion & Loading | MC, LS: AnimatedSection, FadeIn, Skeleton, ProgressBar, StepProgressBar |
| 4 | v1.4.0 Premium UI Complete | PUI, FC: Wire components, Lighthouse ≥90, WCAG AA |
| GitHub Milestone | Status |
|---|---|
| v1.0.0 Core | ✅ Closed |
| v1.2.0 React Core & Migration | Open |
| v1.1.0 Design Tokens & Motion | Open |
| v1.4.0 Premium UI Complete | Open |
| v1.3.0 Motion & Loading States | Open |
npm install
npm start # Run server (foreground)
npm run dev # Run with auto-restart on crash/file changes — keeps localhost upOpen http://localhost:3000. Use Try It to diagnose an incident, or call the API directly.
curl -X POST http://localhost:3000/api/diagnose \
-H "Content-Type: application/json" \
-d '{"incident": "Database query takes 45 seconds, CPU 100% on replicas"}'| Feature | Description |
|---|---|
| 4-agent pipeline | Router → Retriever → Skeptic → Verifier (evidence-first) |
| Single incident | POST /api/diagnose, JSON in, root cause and fix plan out |
| Batch | POST /api/batch-diagnose, up to 100 incidents per request |
| Webhooks | Register a URL; results delivered automatically |
| Export | JSON/CSV for runbooks and compliance |
| Audit trail | Immutable logs with trace IDs |
| Orchestration UI | View tasks, approvals, agents at /orchestration.html |
- Web UI: Go to http://localhost:3000, scroll to Try It, describe the incident, click Diagnose.
- API:
POST /api/diagnosewith{"incident": "..."}(10-2000 chars). - Product section: Scroll to Integrate With Your Stack for tabs: REST API, Batch, Webhooks, Export, All Endpoints.
| Use Case | Example Input | What You Get |
|---|---|---|
| API / Latency | "API latency spiked to 12s in us-east-1 after deploy v2.4.1, connection pool at 98%" | Root cause, fix plan, deployment-related bottlenecks |
| Database | "Database query takes 45 seconds, CPU 100% on read replicas" | Slow-query analysis, replication lag, indexing suggestions |
| Auth / 5xx | "API returns 500 on login, stack trace shows NullPointerException in session service" | Code path trace, error source, fix steps |
| Payments | "Payment processing failing, Stripe webhook timeout, retries exhausted" | Third-party integration diagnosis, retry/recovery plan |
| On-Call / SRE | "Pod restart loop in production, OOMKilled, memory limit 512Mi" | Resource diagnosis, scaling or leak recommendations |
| CI / Flaky Tests | "E2E checkout test fails intermittently, timeout after 30s" | Flakiness vs code bug, timeout/ordering suggestions |
| Microservices | "Service A calls B, B returns 503, circuit breaker open" | Cascade analysis, circuit breaker config, fallback options |
Tip: Include symptoms, timestamps, error messages, and environment (region, version) for better results.
# Single incident
curl -X POST http://localhost:3000/api/diagnose \
-H "Content-Type: application/json" \
-d '{"incident": "API latency spiked to 12s after deploy v2.4.1"}'
# Batch (up to 100)
curl -X POST http://localhost:3000/api/batch-diagnose \
-H "Content-Type: application/json" \
-d '{"incidents": ["...", "..."]}'Every response includes X-Trace-Id for end-to-end tracing.
curl -X POST http://localhost:3000/api/webhooks \
-H "Content-Type: application/json" \
-d '{"url": "https://your-service.com/hooks/diagnosis"}'Pass webhook in the diagnose request to push results to your endpoint.
Use any HTTP client (Python, Node, Go, bash). Set Content-Type: application/json. Full reference: http://localhost:3000/api-reference.html
This repo includes 118+ skills in .claude/skills/ for run-the-business automation, evidence-proof workflows, and ultra-automation. Project 1.0.0 complete; handoff notes in .claude/SKILLSETS.md.
| Doc | Purpose |
|---|---|
| .claude/SKILLSETS.md | Skill index; roles, user prompts, all governance |
| REVIEW.md | Code review guidelines for Claude Code Review |
Use: Open in Claude Code; skills load from .claude/skills/. Invoke with /run-the-business, /plan-and-execute, or any skill name. Default: maximum automation, auto-merge when CI green.
git clone https://github.com/jimmymalhan/codereview-pilot.git
cd codereview-pilot
npm installnpm test # Unit + integration
npm run test:ci # CI mode (GitHub Actions)
npm run test:watch # Watch mode
npm run test:e2e # E2E (requires ANTHROPIC_API_KEY)
npm run load-test # 20 concurrent diagnose requests (requires server running)- Read CLAUDE.md for project standards.
- For skill work: see .claude/SKILLSETS.md and
.claude/skills/. - Create a branch:
git checkout -b feature/your-feature. - Implement changes and add/update tests.
- Run
npm testbefore committing.
- Commit with Conventional Commits (e.g.
feat(api): add X,fix(ui): Y). - Open a PR. CI runs tests on Node 18 and 20.
- Update CHANGELOG.md with what changed and why.
| Area | Examples |
|---|---|
| API | New endpoints, validation, rate-limit tuning |
| UI | Try It form, orchestration dashboard, accessibility |
| Pipeline | Router, Retriever, Skeptic, Verifier logic |
| Skills | New skills in .claude/skills/; index in .claude/SKILLSETS.md |
| Tests | Unit, integration, E2E, rate-limit coverage |
| Docs | API reference, integration guides, examples |
Incident → Router → Retriever → Skeptic → Verifier → Root Cause + Fix Plan + Tests
- Router: Classifies type and severity.
- Retriever: Fetches evidence from codebase/logs.
- Skeptic: Proposes competing theories.
- Verifier: Produces final root cause and confidence (0-100).
| Method | Endpoint | Purpose |
|---|---|---|
| POST | /api/diagnose |
Single incident diagnosis |
| POST | /api/batch-diagnose |
Batch (up to 100) |
| GET | /api/diagnose/:id |
Retrieve by ID |
| GET | /api/diagnose/:id/export |
Export JSON/CSV |
| GET | /api/diagnostics |
List (paginated) |
| GET | /api/audit-log |
Audit trail |
| POST | /api/webhooks |
Register webhook |
| GET | /health |
Health check |
Full docs: api-reference.html
ANTHROPIC_API_KEY=sk-ant-... # Required for diagnosis
PORT=3000 # Default 3000
NODE_ENV=production # OptionalRate limit: 100 requests/hour per IP (configurable).
- App: http://localhost:3000
- API Reference: http://localhost:3000/api-reference.html
- Orchestration: http://localhost:3000/orchestration.html
- Health: http://localhost:3000/health
Awaiting your confirmation to automate the remaining roadmap. Once confirmed:
- Org-chart + org-feedback-loop — 50 roles (Junior→Founder) give critique and pushback; resolve conflicts; implement until 1.0.0 complete
- Run-the-business — Spawn agents; plan-and-execute; PR → consensus → merge; no manual steps
- Until 1.0.0 — Loop: org feedback → implement next batch → test → PR → repeat until all checkpoints done
Reply to confirm and automation will proceed. No further direction needed—agents run until Project 1.0.0 is complete.
Code Review Pilot | Evidence first. Root cause in seconds. MIT