I Want More Coding Agents to Work Like This
💻 One thing I dislike about coding-agent setups is how quickly they become part of one specific machine. Provider config goes in one place, session state somewhere else, local models live in another directory, and suddenly moving to a second machine means rebuilding the environment. OpenClaude-Portable takes a much cleaner approach. It packages the coding agent, runtime and persistent data into…
Coding-agent setups often become tied to a single machine, with configuration, session state, and local models scattered across different locations. OpenClaude-Portable takes a different approach, packaging the coding agent, runtime, and persistent data into a single, self-contained folder. This setup supports both cloud and local models, allowing users to switch between different providers without rebuilding their environment.
Currently, it supports nine providers, including Anthropic Claude, OpenAI, Google Gemini, DeepSeek, OpenRouter, NVIDIA NIM, Ollama, LM Studio, and custom OpenAI-compatible APIs.
The portable nature of OpenClaude-Portable is not tied to a specific model vendor, allowing users to use cloud models when needed and switch to local options like Ollama or LM Studio when working offline. The caveat is that cloud providers still require an internet connection, while Ollama can run locally after the initial setup.
The project redirects persistent data into a local data folder, including provider settings, API keys, logs, session history, agent memory, and local Ollama files. Importantly, it does not write configuration into the host system, which is crucial for maintaining a consistent environment when moving the folder around.
OpenClaude-Portable offers two modes for running the agent: normal mode, which prompts for file writes or shell commands, and limit-free mode, which runs without approval prompts. The project also includes session resume functionality, storing session history inside the portable data folder and providing a resume command for interrupted sessions. This is more valuable than flashy UI features, as it allows the agent's working context to persist over time.
The browser dashboard offers a chat-style interface, agent mode, tool cards, and thinking visualization, providing an alternative to a terminal-only experience. OpenClaude-Portable's local-model optimization is particularly impressive, with a local proxy that trims the system prompt for small Ollama models, reducing latency from 60-120 seconds to roughly 5-20 seconds on CPU-only hardware.
The base installation is small, requiring around 150 MB for the runtime and engine, while local models can grow to 800 MB to 8 GB, depending on the Ollama model used. This makes a portable setup practical for various use cases, such as travel, backup machines, recovery environments, temporary workstations, and offline coding with small local models.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
