Package Exports
This package does not declare an exports field, so the exports above have been automatically detected and optimized by JSPM instead. If any package subpath is missing, it is recommended to post an issue to the original package (@cortiqa/cordenex) to support the "exports" field. If that is not possible, create a JSPM override to customize the exports field for this package.
Readme
🤖 Cordenex v2 — Autonomous AI Software Engineer
Cordenex is a high-performance, autonomous AI coding assistant designed for the terminal. Inspired by the logic of Claude Code and the power of specialized agentic workflows, Cordenex brings Anthropic-level engineering capabilities to your local environment.
"Stop chatting with code. Start engineering with it."
🔥 Key Pillars of Cordenex v2
1. 🧠 Project Intelligence & Memory
CORDENEX.md(Project Memory): Define project-specific rules, conventions, and tech-stack details that persist across every AI turn.- Auto-Context Detection: Cordenex automatically analyzes your project (languages, frameworks, dependencies) to provide hyper-relevant suggestions without you explaining anything.
- Session Journaling: Maintains a rolling "Journal" of what was accomplished, providing continuity even if you restart the terminal.
2. ⚡ Autonomous Execution Engine
- Parallel Tool Execution: Executes multiple file operations, shell commands, and searches concurrently to save time.
- Self-Healing Workflows: If a tool fails (e.g., missing directory, syntax error), Cordenex automatically diagnoses the issue, fixes its own mistakes, and retries.
- Smarter Mode Switching: Automatically switches between Chat Mode (for explanations) and Agent Mode (for complex migrations) based on the task complexity.
3. 🛡️ Safety & Permission Tiers
- Tiered Permissions:
- Read-only: Always allowed (safe).
- Write: Needs approval by default.
- Dangerous: Shell access and deletions always ask for confirmation.
- Smart Approvals: Approve a batch of parallel actions once, not one by one.
4. 🛠️ Developer First Experience
- Rich Interactive UI: Features a streaming interface with real-time Thinking Steps (Reasoning blocks).
- Universal Provider Support: Anthropic (Claude 3.5), OpenAI (GPT-4o), Google (Gemini 2.0), Groq (Llama 3.3 for speed), and Ollama for full offline privacy.
- Undo System: Accidentally broke something? Use
/undoto revert file changes instantly.
🚀 Installation & Setup
1. Install via NPM (Recommended)
npm install -g @cortiqa/cordenex2. Setup your Provider
Cordenex supports all major AI backends. We recommend Claude 3.5 Sonnet (via Anthropic) or Llama 3.3 (via Groq) for the best results.
# Set your API Key
cordenex config set providers.anthropic.api_key your_key_here
# Set default model
cordenex switch claude-3-5-sonnet💻 Commands Reference
| Command | Description |
|---|---|
cordenex chat |
Start the main interactive engineering session. |
cordenex ask "prompt" |
Run a one-shot task and exit. |
cordenex init |
Initialize CORDENEX.md and project rules. |
cordenex auth login |
Authenticate with Cortiqa Cloud. |
cordenex config set <key> <val> |
Configure settings (UI, AI, Logging). |
cordenex models |
List all supported and configured models. |
⚡ Slash Commands (Within Chat)
/undo: Revert the last set of file changes./memory: View or edit your project-specificCORDENEX.md./permissions: Change safety level (allow-write,allow-all,default)./changes: Show a summary of files modified in this session./clear: Reset the conversation history./model <id>: Switch models on the fly.
🏗️ Architecture
Cordenex v2 is built in Go for maximum speed and zero-dependency binaries.
internal/core: The brain. Contains theEngine(LLM logic),Orchestrator(mode switching), andJournal(memory).internal/tools: The hands. Filesystem operations, Shell execution, and Context retrieval.internal/providers: The interface. Unified API for OpenAI, Anthropic, Gemini, etc.
For deeper technical details, see ARCHITECTURE.md.
🤝 Contributing
We welcome contributions!
- Fork the repo.
- Run
go build ./cmd/cordenexto test changes. - Submit a PR.