Deep Coder is a project-scoped terminal coding agent built around the DeepSeek API.
# Deep Coder Deep Coder is a project-scoped terminal coding agent built around the DeepSeek API. It launches as a Textual TUI, keeps session history isolated per workspace, streams tool activity into a replayable timeline, and persists runtime state under `~/.deepcode/`. The current product is the package-based runtime under `deep_coder/` plus the `deepcode` launcher. `agentLoop.py` remains in the repository as a legacy prototype and reference file, not the main entrypoint. ## Highlights - Project-scoped sessions: each workspace resolves to a stable project record and its own persisted state root. - Terminal-first workflow: one timeline, one composer, live event streaming, and replay of stored sessions. - Local coding tools: bash, file read/write/edit, session task tools, and layered history search/load tools. - Layered context: recent turns, persisted evidence, and summaries are assembled through the context layer instead of a flat transcript replay. - DeepSeek-backed runtime: the model adapter uses the OpenAI-compatible Python SDK with DeepSeek as the current provider. ## Installation ### Install From Source As A Command Clone the repository and install it into your user site-packages so `deepcode` is available on your shell `PATH`: ```bash git clone <repository-url> cd Deep-Coder python3 -m pip install --user . ``` On many Linux systems this installs the command into `~/.local/bin`. If `deepcode` is not found after installation, add that directory to your `PATH`. ### Editable Install For Contributors If you want an editable checkout for development: ```bash git clone <repository-url> cd Deep-Coder python3 -m venv .venv . .venv/bin/activate python3 -m pip install -e ".[dev]" ``` That exposes `deepcode` inside the virtual environment and installs `pytest` for local verification. ### Required Environment Deep Coder currently requires a Claude API key: ```bash export DEEPSEEK_API_KEY="your-api-key" ``` ### Launcher Fallback For repository-local de
HAL 分层混合模型工作流 — 强模型(Claude)负责理解/拆解/验收,低成本模型(DeepSeek)负责检索/提取/清洗。Hermes Agent skill。
An LLM agent fine-tuned on DeepSeek for spaced repetition, dynamically integrating knowledge points based on the Ebbinghaus forgetting curve.
基于 STM32F103 构建的端到端 AI 智能手表生态。自研“零重定位”原生机器码动态加载引擎与页面栈式 UI 框架;集成生产级 OTA 回滚保护机制与高带宽(921600 baud)串口协议栈。通过 Node.js 中继实现 DeepSeek AI 语义控制及 ASRPRO 语音全双工交互,是一个集成了分布式计算、现代存储管理与 AI Agent 的嵌入式全栈工程。
A Meta-Agent-Driven Self-Evolving Multi-Agent System for UAV Detection and Tracking
One command to run Hermes AI Agent with a browser UI. Zero prerequisites. 一行命令,AI 就位。
网页应用Agent,接入DeepSeek、Mimo等模型