AI coding agent CLI, built on VS Code architecture with multi-provider LLM support and Cursor-compatible skills.
- Multi-mode: Agent (full autonomy), Plan (implementation planning), Ask (read-only Q&A)
- Multi-provider: OpenAI, Anthropic, Ollama, and any OpenAI-compatible API (DeepSeek, etc.)
- YAML config with profiles: No more juggling environment variables
- Cursor-compatible Skills: Load SKILL.md and .mdc rules to extend agent capabilities
- Streaming output: Real-time token display
- Parallel agents: Run multiple tasks concurrently
- Single-file distribution: 70 KB minified, zero runtime dependencies
# Install from GitHub Releases
npm install -g https://github.com/zhouronghua/code-agent/releases/latest/download/code-agent-0.3.10.tgz
# Or install from npm (if published)
npm install -g code-agent
# Create config
mkdir -p ~/.codeagent
cp $(npm root -g)/code-agent/config.template.yaml ~/.codeagent/config.yaml
# Edit config.yaml: fill in your API key
# Run
code-agent "write a hello world in Python"
code-agent --mode plan "design a REST API"
code-agent --mode ask "explain the project structure"code-agent [options] "task description"
Options:
--mode <agent|plan|ask> Agent mode (default: agent)
--stream Streaming output
--parallel "t1" "t2" Run tasks in parallel
--profile <name> Use a config profile
--profiles List available profiles
--skills List loaded skills
--use-skill <name> Activate a skill for this session
--help Show help
See USAGE.md for full documentation including configuration, skills, distribution, and more.
src/vs/workbench/
services/agent/ # Core models, service interface, LLM providers
contrib/agent/ # Agent loop, tools, context, skills, config
vs-core/
base/common/ # VS Code API shims (event, lifecycle, uri, etc.)
platform/ # Service interfaces (file, search, terminal)
node-runtime/ # Node.js service implementations + CLI entry
Built with TypeScript, bundled with esbuild into a single executable JS file.
npm install
npm run build # Development build
npm run build:release # Production build (minified)
npm run pack # Create distributable .tgzMIT