appifex-dtc

Design to TestFlight in one command.

dtc is a CLI that turns design files (Pencil / Figma / Stitch) into production-ready native and web apps deployed to TestFlight or the Play Store — no manual Xcode, Android Studio, or Firebase Console steps required.

Status: v1.0 MVP — SwiftUI + Firebase path is stable and end-to-end tested. Kotlin Compose, React, and React Native targets are in active development.

Features

• Design → code in one command — feed a .pen, Figma URL, or Stitch file; get a compilable, linted SwiftUI or Kotlin Compose project back
• Layered codegen pipeline — design extraction → spec generation → test generation → codegen → build → validation → LLM-assisted auto-fix → delivery
• BaaS wiring — Firebase (Firestore, Auth, Storage) and Supabase schemas inferred from your design and scaffolded automatically
• Validation gate — Maestro UI tests + semgrep security scan block delivery on hard failures; xcodebuild archive confirms the binary actually compiles
• Auto-fix loop — on validation failure the pipeline feeds errors back to the LLM within a tracked TokenBudget and retries up to a configurable limit
• Resume / checkpoint — every phase is persisted to SQLite; interrupted runs pick up where they left off
• MCP server — dtc-mcp-server exposes the full pipeline as MCP tools so AI coding agents (Claude, Cursor, etc.) can orchestrate generation workflows
• Multi-provider LLM — Anthropic Claude (SDK + claude --print), GitHub Copilot, OpenAI, and Google Gemini all supported

Supported platforms

Install

npm install -g @appifex/cli
# or
pnpm add -g @appifex/cli

Requires Node ≥22, pnpm ≥9, macOS with Xcode 15+ (for iOS/Swift targets).

Quickstart

# One-time setup: LLM API key, Apple ASC credentials, Firebase project
dtc setup

# Run the full pipeline
dtc run --design my-app.pen --platform swiftui

# → Extracts design tokens and component tree
# → Generates SwiftUI project + Firebase rules + Firestore schema
# → Archives with xcodebuild
# → Runs Maestro UI tests + semgrep security scan
# → Uploads to TestFlight via App Store Connect API

See docs/getting-started.md for the complete walkthrough including credential setup and troubleshooting.

MCP server

dtc ships an MCP server that AI coding agents can use to drive the pipeline:

# Add to your agent's MCP config
dtc-mcp-server

The server exposes pipeline stages as discrete MCP tools — useful for agents that want to run only codegen, only validation, or only delivery.

How it works

Design file (Pencil / Figma / Stitch)
    │
    ▼ design extraction
PlatformSpec + DesignTokens
    │
    ▼ spec generation
Component tree + BaaS schema
    │
    ▼ test generation
Maestro flow + unit test stubs
    │
    ▼ codegen (LLM)
SwiftUI / Kotlin Compose source
    │
    ▼ build (xcodebuild / Gradle)
Compiled binary
    │
    ▼ validation (Maestro + semgrep)
Pass / Fail → auto-fix loop if needed
    │
    ▼ delivery (gh + ASC API)
TestFlight build

Each stage is a checkpointed phase. A failed or interrupted run resumes from the last successful phase.

CLI reference

dtc setup              One-time credential and project configuration
dtc run                Execute the full pipeline (see dtc run --help)
dtc doctor             Check prerequisites and credential health
dtc resume             Resume an interrupted pipeline run
dtc report             Print the last run report

See docs/cli-reference.md for all flags and options.

Repository layout

cli/                   CLI entry point and pipeline orchestration
packages/
  core/                Domain types, config, checkpoint, token budget
  design/              Design file adapters (Pencil, Figma, Stitch)
  spec/                Platform spec translation
  codegen/             LLM codegen (Claude, Copilot, OpenAI, Gemini)
  test-gen/            Test scaffold generation
  build/               xcodebuild / Gradle wrappers
  validate/            Maestro UI tests + semgrep security scan
  fix/                 LLM-assisted auto-fix loop
  baas/                Firebase / Supabase schema inference + templates
  provision/           Apple ASC + Google Play provisioning
  report/              Run report generation
  runner/              Local / E2B / remote execution adapters
  mcp-server/          MCP server exposing pipeline as tools
  agent/               LLM agent adapters (Claude, Codex, Gemini)

Documentation

• Getting Started
• CLI Reference
• Contributing
• Security Policy

Contributing

See CONTRIBUTING.md. This project is maintained by Appifex AI Technologies, Inc. — PRs are welcome, but review throughput is limited.

License

MIT — see LICENSE.