mereyabdenbekuly-ctrl/clodex-ide
Local-first, zero-trust agentic IDE for verifiable autonomous software development.
What it solves
Most AI coding interfaces focus on single-message interactions, which are insufficient for complex engineering work. Real development requires understanding entire repositories, planning multi-file changes, running commands, and inspecting results over long periods. CLODEx addresses this by treating engineering work as a durable, continuous task rather than a series of disposable chat messages.
How it works
CLODEx provides a unified workspace that integrates files, Git, a terminal, a browser, and MCP (Model Context Protocol) tools. It uses a local-first approach to keep task history and workspace state persistent, allowing users to resume work across sessions without rebuilding context. Users can connect to models via CLODEx accounts, personal API keys, custom OpenAI-compatible endpoints, or local providers like Ollama. The system emphasizes human oversight through explicit approval surfaces for sensitive actions, line-level diffs, and pending edit reviews.
Who it’s for
Software engineers and developers who need an agentic development environment capable of handling long-running, complex tasks while maintaining full control and visibility over AI-generated changes.
Highlights
- Durable tasks: Maintains searchable history and workspace context across restarts.
- Unified workspace: Integrates Git, terminal, browser, and MCP tools in one place.
- Flexible model support: Supports BYOK (Bring Your Own Key), custom endpoints, and local Ollama.
- Review-centric design: Features explicit permission checks and line-level diffs for all agent actions.