AI-powered project orchestration platform. Describe a project in natural language — CreedFlow analyzes it, creates tasks, routes them to cloud or local AI backends, reviews code, generates creative assets, publishes content, and deploys.
 |
 |
| Project Dashboard — Overview with task stats and quick actions | Kanban Board — Tasks grouped by status with live agent output |
 |
 |
| Agents Working — Real-time orchestration with backend routing | Task Details — Full task list with code review panel |
 |
 |
| Setup Wizard — Auto-detect AI CLIs with version info | Dependencies — One-click install via Homebrew |
 |
 |
| Agent Preferences — Per-agent backend routing config | MCP Servers — 13 integrations (DALL-E, Figma, Runway...) |
| Platform | Architecture | Version | Download |
|---|---|---|---|
| macOS | Apple Silicon (M1/M2/M3/M4) | v1.3.0 | Download DMG |
| macOS | Intel | v1.3.0 | Download DMG |
| Linux | x86_64 (AppImage) | v1.3.0 | Download AppImage |
| Linux | x86_64 (Debian/Ubuntu) | v1.3.0 | Download .deb |
Requirements: macOS 14+ or Linux (Ubuntu 22.04+, Debian 12+). At least one AI backend: Claude CLI, Codex CLI, Gemini CLI, OpenCode, OpenClaw, Ollama, LM Studio, llama.cpp, or MLX.
Since CreedFlow is not signed with an Apple Developer ID, macOS Gatekeeper will show a warning on first launch:
Option A — Right-click:
- Right-click (or Control-click) on
CreedFlow.appin Applications - Select Open from the context menu
- Click Open in the dialog
Option B — Terminal:
xattr -cr /Applications/CreedFlow.appOption C — System Settings:
- Open System Settings → Privacy & Security
- Scroll down to find the CreedFlow blocked message
- Click Open Anyway
This only needs to be done once.
AppImage:
chmod +x CreedFlow_1.2.0_amd64.AppImage
./CreedFlow_1.2.0_amd64.AppImageDebian/Ubuntu:
sudo dpkg -i CreedFlow_1.2.0_amd64.deb- Project Chat Panel — Slide-in chat panel for AI-assisted task planning and brainstorming
- Task Proposals — AI suggests features and tasks inline; approve or reject with one click
- Streaming Responses — Real-time typing indicator with partial content display
- 9 AI Backends — OpenCode and OpenClaw CLI support with auto-detection and smart routing
- Backend count: 7 → 9 (Claude, Codex, Gemini, OpenCode, OpenClaw, Ollama, LM Studio, llama.cpp, MLX)
- Import Existing Projects — Point to an existing directory instead of creating a new one
- Project Creation Wizard — Step-by-step guided project setup with tech stack detection
- Project Docs Export — Bundle architecture docs, diagrams, and README into a single file (for NotebookLM or other tools)
- Project-Type-Aware Analysis — Analyzer produces specialized output per project type (software, content, image, video)
- Prompt Import/Export — Share prompts as JSON files across teams
- Version Diff — Side-by-side comparison of prompt versions with line-level diff
- Prompt Recommender — AI-powered prompt suggestions based on success rate and review scores
- CLI Usage Tracking — Real-time API usage monitoring via Anthropic/OpenAI admin APIs with sidebar progress bars
- MCP Requirements Checker — Auto-detect missing MCP servers based on project type and agent configuration
- Creative AI Services — HeyGen, Replicate, Leonardo.AI MCP templates for video/image generation
- Skill Persona — Assign personality/expertise profiles to tasks for specialized agent behavior
You: "Todo app with React + Node.js + SQLite"
↓
Analyzer → Architecture docs, ER diagrams, data models, task breakdown
↓
Router → Routes tasks to Claude / Codex / Gemini (+ local LLM fallback)
↓
Coder → Writes code, opens feature branches per task
↓
Reviewer → AI code review with 0-10 scoring
↓
Creative agents → Generate images, videos, designs, documents
↓
Publisher → Distributes content to Medium, WordPress, Twitter, LinkedIn
↓
Telegram notification → You approve → Deploy
- 11 AI Agents — Analyzer, Coder, Reviewer, Tester, DevOps, Monitor, ContentWriter, Designer, ImageGenerator, VideoEditor, Publisher
- Deep Analysis — Architecture docs, data models with field-level detail, Mermaid diagrams (ER, flowchart, sequence, class), tasks with acceptance criteria and file lists
- 9 AI Backends — Claude, Codex, Gemini, OpenCode, OpenClaw (cloud) + Ollama, LM Studio, llama.cpp, MLX (local) with smart routing and automatic fallback
- AI Chat — Project-scoped chat panel for AI-assisted planning with inline task proposals
- Kanban Board — Drag-and-drop task management with live agent output
- Project Wizard — Step-by-step project creation with import existing directory support
- Setup Wizard — Environment detection, one-click dependency install via Homebrew (macOS) or apt/dnf/pacman (Linux)
- Asset Pipeline — Creative agents produce images/videos/designs with versioning, checksums, and format variants (.md → .html, .txt, .pdf)
- Content Publishing — Publish to Medium, WordPress, Twitter, LinkedIn with scheduled publishing
- Prompt Library — Versioning, chaining, tagging, import/export, version diff, AI-powered recommendations
- CLI Usage Tracking — Real-time API usage monitoring via Anthropic/OpenAI admin APIs
- Git Integration — Feature branches, auto-commit, auto-merge on review pass, three-branch progression (dev → staging → main)
- Local Deploy — Docker, Docker Compose, or direct process execution
- MCP Server — 13 tools + 5 resources via
creedflow://URIs - Creative MCP — DALL-E, Figma, Stability AI, ElevenLabs, Runway, HeyGen, Replicate, Leonardo.AI integrations
- Telegram Notifications — Task completion, review results, deploy status
| Component | macOS | Linux |
|---|---|---|
| Language | Swift 6.0 | Rust + TypeScript |
| UI | SwiftUI | React + Tailwind CSS (Tauri) |
| Database | SQLite via GRDB.swift | SQLite via rusqlite |
| AI Backends | Claude CLI, Codex CLI, Gemini CLI, OpenCode, OpenClaw + Ollama, LM Studio, llama.cpp, MLX | Same |
| MCP | modelcontextprotocol/swift-sdk | — |
| Deployment | Docker / Docker Compose / Direct Process | Same |
| Notifications | Telegram Bot API | Same |
macOS (Swift):
cd CreedFlow && swift build
.build/debug/CreedFlow
# Package as DMG
./Scripts/package-app.sh --dmgLinux (Tauri):
cd creedflow-desktop
pnpm install
pnpm tauri dev # Development
pnpm tauri build # Production (.deb + .AppImage)┌─────────────────────────────────────────────────────────────┐
│ Desktop App (SwiftUI / Tauri+React) │
│ ┌──────────┐ ┌──────────────┐ ┌────────────────────────┐│
│ │ Sidebar │ │ Task Board │ │ Detail Panel (Right) ││
│ │ Projects │ │ (Kanban) │ │ Live Output / Review ││
│ └──────────┘ └──────────────┘ └────────────────────────┘│
└──────────────────────────┬──────────────────────────────────┘
│
┌──────────────────────────▼──────────────────────────────────┐
│ Engine (Orchestrator) │
│ ┌─────────────┐ ┌──────────────┐ ┌────────────────────┐ │
│ │ Task Queue │ │ Backend │ │ Agent Scheduler │ │
│ │ (Priority + │ │ Router │ │ (Concurrency) │ │
│ │ Dependencies)│ │ (Smart │ │ │ │
│ │ │ │ Fallback) │ │ │ │
│ └─────────────┘ └──────────────┘ └────────────────────┘ │
└──────────────────────────┬──────────────────────────────────┘
│
┌──────────────────────────▼──────────────────────────────────┐
│ AI Backends │
│ │
│ Cloud: Claude CLI · Codex CLI · Gemini CLI · OpenCode · OpenClaw │
│ Local: Ollama · LM Studio · llama.cpp · MLX │
└──────────────────────────┬──────────────────────────────────┘
│
┌──────────────────────────▼──────────────────────────────────┐
│ 11 AI Agents │
│ │
│ Analyzer → Coder → Reviewer → Tester → DevOps → Monitor │
│ ContentWriter → Designer → ImageGenerator → VideoEditor │
│ Publisher │
└──────────────────────────┬──────────────────────────────────┘
│
┌──────────────────────────▼──────────────────────────────────┐
│ Storage & Services │
│ SQLite (GRDB/rusqlite) · Asset Pipeline · Content Publishing│
│ Telegram Bot · MCP Server · Git Branch Manager │
└─────────────────────────────────────────────────────────────┘
| Agent | Backend | Timeout | Purpose |
|---|---|---|---|
| Analyzer | Any | 5 min | Architecture analysis, data models, diagrams, task decomposition |
| Coder | Claude preferred | 15 min | Write code, create branches/PRs |
| Reviewer | Claude preferred | 5 min | AI code review with 0-10 scoring |
| Tester | Claude preferred | 10 min | Generate and run tests |
| DevOps | Any | 10 min | Docker, CI/CD, infrastructure setup |
| Monitor | Any | 5 min | Health checks, log analysis |
| ContentWriter | Claude preferred | 10 min | Articles, docs, copy writing + multi-format export |
| Designer | Claude preferred | 10 min | Design specs + Figma access |
| ImageGenerator | Claude preferred | 10 min | AI image generation (DALL-E, Stability) |
| VideoEditor | Claude preferred | 15 min | Video/audio generation (Runway, ElevenLabs) |
| Publisher | Claude preferred | 10 min | Medium, WordPress, Twitter, LinkedIn |
CreedFlow also runs as an MCP server (CreedFlowMCPServer binary):
Tools (13): create-project, create-task, update-task-status, get-project-tasks, run-analyzer, get-cost-summary, search-prompts, list-assets, get-asset, list-asset-versions, approve-asset, list-publications, list-publishing-channels
Resources (5): creedflow://projects, creedflow://tasks/queue, creedflow://costs/summary, creedflow://projects/{id}/assets, creedflow://publications
CreedFlow orchestrates AI agents that generate and modify code autonomously. When using AI-powered code generation, follow these security practices:
- Review all AI-generated code before committing or deploying. AI agents can produce code with unintended vulnerabilities.
- Never commit secrets — API keys, tokens, passwords, and private keys should never be stored in your repository. Use environment variables or a secrets manager.
- Use
.gitignore— Ensure.env,*.pem,*.key, and other sensitive file patterns are excluded from version control. - Sandbox your deployments — CreedFlow supports Docker-based deployment. Use containers to isolate AI-generated code from your host system.
- Monitor agent output — Review agent logs and generated files before approving tasks, especially for DevOps and deployment operations.
- Keep backends updated — Regularly update your AI CLI tools (Claude, Codex, Gemini, etc.) to get the latest security patches.
CreedFlow stores API keys and tokens locally on your machine (UserDefaults on macOS, settings.json on Linux). These are never transmitted to third parties beyond the configured AI backends.
If you find CreedFlow useful, consider supporting the project:
MIT — Copyright (c) 2025 Fatih Kan