Deep Code CLI
A terminal coding agent tuned for the deepseek-v4 model: deep thinking, reasoning effort control, Agent Skills and MCP
High risk
We rate an entry high when the tool writes to external systems, handles money, production databases or secrets, or runs arbitrary commands. The CLI installs it only with your consent.
Why this level
- Runs shell commands and edits project files as a coding agent
- Sends project code to an external model API under the user's key
Install
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/deep-code-cliDetects the agents on your machine, checks the risk and pins the version.
Other ways to install
Install the tool
npm install -g @vegamo/deepcode-cliInstall globally: npm install -g @vegamo/deepcode-cli. Create ~/.deepcode/settings.json with env.MODEL, env.BASE_URL and env.API_KEY for your provider, then run deepcode in the project directory.
Other ways from the author
npm install -g @vegamo/deepcode-cliAfter installing, create ~/.deepcode/settings.json with the model, BASE_URL and provider key, then run deepcode in the project directory. Node.js 22 or newer is required.
This is third-party code. Review the repository files before installing.
What it does
Deep Code is a terminal coding assistant that the author deliberately tuned for DeepSeek models so that the shape of its tools matches how those models were trained to call them. It supports deep thinking and reasoning effort control, context caching to cut cost, and loading Agent Skills and external tools over MCP. The interface has a plan mode, conversation forking and continuation, undo for code and chat, model switching and an MCP status view. The model can read local images and images from the clipboard, and it can call a notification script when a task finishes. Its configuration is shared with the Deep Code VSCode extension, so settings carry over between the terminal and the editor.
Who it is for. For developers who want to code in the terminal on DeepSeek models or another OpenAI-compatible provider.
Good fit when
- You need a terminal coding agent tuned for DeepSeek
- You want to control reasoning depth and save cost with context caching
- You need Agent Skills and external tools over MCP in a console workflow
Not a fit when
- You work only in a GUI and do not need a terminal interface
- You have no access to the DeepSeek API or another OpenAI-compatible endpoint with a key
Example request
Get familiar with this project and add error handling to the file upload moduleLimitations
It needs a key for the DeepSeek API or another OpenAI-compatible provider, and access is paid. This is a coding agent: it runs shell commands and reads and edits project files, so grant access with the permission policy in mind. The project is optimized for DeepSeek, and behavior on other models may differ. Direct image understanding is available on the deepseek-flash model; other models use a separate tool. Node.js 22 or newer is required.
How to disable. Remove the package with npm uninstall -g @vegamo/deepcode-cli. Settings and data live in the ~/.deepcode/ directory.
Security check
- Runs shell commands and edits project files as a coding agent
- Sends project code to an external model API under the user's key
README in short
The README presents Deep Code as a terminal coding assistant for the deepseek-v4 model with deep thinking, reasoning effort control and support for Agent Skills and MCP. Installation is via npm, settings live in ~/.deepcode/settings.json and are shared with the VSCode extension. Skills are discovered from project and user directories, including the interoperable .agents/skills folder. It documents a set of slash commands and shortcuts, a list of supported DeepSeek models and work with images, web search and notifications. The author cites the idea that tool schemas are not neutral, so the harness is aligned to DeepSeek's behavior. MIT licensed.
FAQ
Is there a VSCode extension?
Yes, Deep Code has a VSCode extension that shares the single ~/.deepcode/settings.json config with the CLI, so you can switch between the terminal and the editor.
Can I use something other than DeepSeek?
Yes, any OpenAI-compatible model works via the BASE_URL and API_KEY settings, but the project is optimized for DeepSeek.
Related
A skills library that gives coding agents a development process: brainstorming, planning, TDD, subagents and code review
Skills for real engineers by Matt Pocock
Skills For Real Engineers
Small composable skills for engineering with agents: plan grilling, TDD, bug diagnosis, code review and architecture
GitHub toolkit for spec-driven development: the specify CLI adds agent commands and skills to a project, from principles to implementation
Reference MCP servers
Model Context Protocol servers
Official reference MCP servers: Filesystem, Fetch, Git, Memory, Sequential Thinking, Time and Everything