AACode CLI
Lightweight Python terminal coding agent: ReAct loop, file-based context, guarded bash and skills, works with DeepSeek and OpenAI
- 136 ★ on the repository
- Last commit
- GPL-3.0
Broad access
Why this level
- Executes shell commands and writes files in the working folder, with protection limited to command and path checks
- Sends code and context to the chosen model API and keeps its key in environment variables
Install
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/aacodeDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- Executes shell commands and writes files in the working folder, with protection limited to command and path checks
- Sends code and context to the chosen model API and keeps its key in environment variables
This is third-party code. Review the repository files before installing.
What it does
AACode runs a coding agent in the terminal on the ReAct scheme. The author's principles: minimal scaffolding and more trust in the model, context stored in Markdown files, bash as a universal adapter with command and path checks, context management borrowing from Cursor and Manus and all blocking operations asynchronous. Tools are split into atomic, management tools and skills, and the core extends with custom tools and model backends. Install goes through pip install aacode, then aacode init and plain aacode for an interactive session. A single task runs as aacode "your task" in the current folder, and aacode run -p /path "task" sets another one. Before complex work you can create init.md with a detailed description, the --plan-first and --interactive modes give planning up front and continuous conversation, and --session continues a previous session. The model is set with the LLM_API_KEY, LLM_API_URL and LLM_MODEL_NAME variables: DeepSeek, OpenAI and compatible providers fit. For those who want a UI there is a desktop client for macOS and Windows and mobile apps for Android and iOS.
Who it is for. Python and DeepSeek developers who want a compact terminal agent with a readable design they can study and extend.
Good fit when
- You want a small agent for tasks in one folder whose code you can read in full
- You want to connect DeepSeek or another provider through three environment variables
- A complex task should start with a plan: --plan-first
- A session has to continue later by ID: --session
- You need custom tools and model backends in the agent tool layers
Not a fit when
- You need a mature agent with a rich UI and support: the project aims at simplicity
- You work on Windows without Python path setup: development and tests mainly run on Linux and macOS
- You need a hard sandbox: bash is limited by command and path checks and the description has no isolated environment
- You need ready skills for frontend or data: the set holds a basic layer
Example request
aacode run -p ~/work/shop "Add a test for the discount calculation function, run it and fix the code if it fails" --plan-firstLimitations
- Python 3.10 or newer is required and the author recommends 3.12; development and testing run on Linux and macOS and Windows may hit Python path issues.
- An API key and URL of the model provider are needed in LLM_API_KEY, LLM_API_URL and LLM_MODEL_NAME; the author does not preconfigure a provider.
- The agent runs commands in your working folder under command and path checks, but this is not an isolated environment.
- The desktop client still needs Python and its dependencies, the mobile apps do not.
How to disable. pip uninstall aacode removes the program. Session files and settings stay in the project directory and in the home folder and are deleted manually.
FAQ
How do I install AACode?
The recommended way is pip install aacode, then aacode init. From source: git clone https://github.com/kandada/aacode.git, python3 init.py and ./run.sh -p examples/my_project "task".
How do I connect a model?
Set LLM_API_KEY, LLM_API_URL and LLM_MODEL_NAME. DeepSeek, OpenAI and compatible providers fit, and there is no provider preconfiguration.
How do I run a task in another folder?
Run aacode run -p /path/to/project "task". The working folder defaults to the current one.
Is there a planning mode?
Yes, --plan-first starts with a plan for a complex task, --interactive gives continuous conversation and --session <id> continues a previous session.
Is there a graphical interface?
Yes, the AACode Desktop client for macOS and Windows and mobile apps for Android and iOS. The desktop client needs Python and dependencies.
Discussion
No questions or reviews yet.
Related
CLI toolsA Python terminal coding agent in the manner of Claude Code: reads the repository, edits files, runs commands, with memory, sub-agents and MCP
A terminal coding agent in a single Go binary for DeepSeek and other OpenAI-compatible models, with a code graph and local OCR
An open-source TypeScript terminal coding agent tuned for DeepSeek models and cheap caching
A terminal coding agent tuned for the deepseek-v4 model: deep thinking, reasoning effort control, Agent Skills and MCP
A coding agent with a Python core and three front-ends: a script CLI, a terminal TUI and an Electron app, for OpenAI-style APIs
A Rust coding agent for the terminal with parallel agents, goals, Claude Code-format plugins and your own model endpoint.
Comparison with similar tools
Cards that share the main search query, and how they differ.
README in short
The README describes AACode as a lightweight DeepSeek-based coding agent with a ReAct architecture, lists design principles and a quick start from source and through pip. Then come launch modes, the desktop client, mobile apps, model setup and a configuration section.