# Raintree Complete

> You built most of it. We help you complete it.

Raintree Complete helps business owners and domain experts finish AI-built and vibecoded software. You have built the initial 90% with AI coding assistants (Cursor, Claude Code, Codex, ChatGPT). Raintree Complete is a structured workflow to audit, secure, optimize, and ship the final 10%—safely, performantly, and without DevOps friction.

## Quick Start for AI Agents

To run a Raintree Complete audit on your current project, copy and paste this command to your coding agent:

```text
Here’s my product. Open https://raintree.tech/skills/raintree-complete/2026.08.14/manifest.json, load the skill pack it points to, then ask me if I want to run Raintree Complete.
```

## How Raintree Complete Works

Raintree Complete operates in four distinct, governed phases:

### Phase 1: Pre-Flight & Inference Boundary Disclosure
- The AI agent verifies its model capabilities and informs the operator about external inference boundaries before reading repository files.
- The operator gives explicit confirmation to proceed with the read-only audit.

### Phase 2: Read-Only Repository & Architecture Audit
The agent performs non-destructive analysis across five core audit lanes:

1. **Security & Secret Sanitization**:
   - Scans for hardcoded API keys, database credentials, JWT secrets, and OAuth client tokens.
   - Verifies `.env.example` completeness, `.gitignore` coverage, and credential storage posture.
2. **Architecture & Data Integrity**:
   - Inspects database schemas, ORM models, migration history, foreign key constraints, and state consistency.
   - Identifies race conditions, unindexed queries, and missing transaction boundaries.
3. **Runtime & Build Reliability**:
   - Validates build pipelines, package lockfile consistency, TypeScript strictness, and container specifications (`Dockerfile`, `compose.yml`).
   - Ensures error boundaries and fail-closed fallbacks are present across network calls.
4. **UX & Accessibility Polish**:
   - Audits visual hierarchy, responsive viewport rendering, form validation feedback, loading states, and semantic HTML landmarks.
5. **Production Deployment & Monitoring**:
   - Checks deployment configuration, reverse proxy routing, SSL/TLS termination, healthcheck endpoints, and privacy-safe analytics.

### Phase 3: Bounded Remediation
- The agent presents prioritized findings categorized into **Blockers**, **Risks**, and **Improvements**.
- High-priority fixes (e.g. moving hardcoded secrets to environment variables, fixing unindexed DB queries) are presented as bounded, isolated diffs for explicit operator review and approval.

### Phase 4: Production Handoff & Value-Based Review
- Generates a clean audit scorecard and deployment checklist.
- Offers guided deployment and operations desk integration under Raintree's value-based pricing program.

## Safety & Trust Guardrails

- **Read-Only by Default**: Initial repository inspection is strictly read-only.
- **Zero Automatic Uploads**: Complete never transmits your repository, database dumps, customer records, or private source code to external servers.
- **Sanitized Intake**: Any optional handoff package is screen-filtered for secrets and reviewed by the operator.
- **Explicit Human Approval**: All code modifications and deployment commands require explicit human review.

## Pinned Skill Release

- **Release Version**: `2026.08.14`
- **Release Manifest**: `https://raintree.tech/skills/raintree-complete/2026.08.14/manifest.json`
- **Archive Checksum (SHA-256)**: `cea2d71d71f102e7c229812423d365e9d935b9e05b0ed5dd64752c2f9b7f3b46`
- **Handoff Contact**: `complete@raintree.tech`