A powerful AI-powered DevOps assistant that helps with infrastructure automation, CI/CD pipelines, containerization, and cloud deployments.
- AI-Powered Assistant: Leverages multiple LLMs (Ollama, Google Gemini, Azure OpenAI, OpenAI, Anthropic) for intelligent DevOps guidance
- Local Model Support: Run fully offline with Ollama — no API key required
- Tool Integration: Extensible tool system for executing system commands
- Ink Terminal UI: Full-screen terminal interface with transcript, side panels, status, tools, and a multiline composer
- Rich Terminal Output: Markdown rendering with ANSI colors and formatting
- DevOps Expertise: Specialized in infrastructure automation and deployment strategies
- TypeScript: Fully typed codebase for better development experience
- Modular Architecture: Clean separation of concerns with pluggable providers and tools
- Bun 1.x
- At least one AI provider:
- Ollama running locally (no API key required — the default), or
- Google Gemini API key, or
- Azure OpenAI service endpoint and API key, or
- OpenAI API key, or
- Anthropic API key
-
Clone the repository:
git clone <repository-url> cd devops-agent
-
Install dependencies:
bun install
-
Configure the application:
cp config.yaml.example config.yaml
Edit
config.yamland configure your preferred AI provider.
Run the application in watch mode with auto-reload:
bun devBuild a standalone binary and run it:
bun build
./devopsagentOr run directly without building:
bun startOnce the application is running, you can use the following commands:
/helporhelp— Show available commandsask <question>— Ask the AI agent a specific question/toolsortools— List available tools/statusorstatus— Show AI provider status/clearorclear— Clear conversation history and the transcript/exitorexit— Exit the application
The composer supports multiline prompts. Press Enter to submit, Ctrl+J or Shift+Enter to insert a newline, Ctrl+P to cycle the side panel, and Esc to clear the composer.
ask How do I create a Dockerfile for a Node.js application?
What are the best practices for Kubernetes deployments?
/tools
/clear
/exit
The DevOps AI Agent features rich terminal output with markdown rendering:
- Headings: Color-coded headings with different levels (H1-H6)
- Code Blocks: Syntax-highlighted code blocks with background colors
- Inline Code: Highlighted inline code with cyan background
- Lists: Properly formatted bullet points and numbered lists
- Links: Clickable links with underlines and URLs
- Emphasis: Bold text and italic text formatting
- Tables: Beautiful table formatting with borders
- Install Ollama from ollama.com
- Pull a model:
ollama pull llama3.3(or any supported model) - Ollama runs at
http://localhost:11434by default — no configuration needed beyond specifying the model name
- Visit Google AI Studio
- Sign in with your Google account
- Click "Create API Key" and copy it
- Visit OpenAI's API platform
- Sign in or create an account
- Navigate to API Keys → "Create new secret key"
- Create an Azure OpenAI resource in the Azure Portal
- Deploy a model (e.g., GPT-4o)
- Get your endpoint URL and API key from the resource overview
- Note your deployment name
- Visit Anthropic Console
- Sign in or create an account
- Navigate to API Keys → "Create Key"
The application uses a YAML configuration file (config.yaml). A minimal configuration using Ollama looks like:
default_provider: "ollama"
providers:
ollama:
enabled: true
model: "llama3.3"
base_url: "http://localhost:11434" # Optional, this is the defaultA full configuration showing all available providers:
default_provider: "ollama"
providers:
ollama:
enabled: true
model: "llama3.3"
base_url: "http://localhost:11434" # Optional
temperature: 0.7 # Optional, 0.0–2.0
gemini:
enabled: false
api_key: "your-gemini-api-key"
model: "gemini-2.5-flash"
temperature: 0.7 # Optional, 0.0–2.0
max_tokens: 4096 # Optional
azure_openai:
enabled: false
api_key: "your-azure-api-key"
endpoint: "https://your-resource.openai.azure.com"
deployment_name: "your-deployment"
api_version: "2024-02-15-preview"
temperature: 0.7 # Optional, 0.0–2.0
max_tokens: 4096 # Optional
openai:
enabled: false
api_key: "your-openai-api-key"
model: "gpt-4o"
organization: "your-org-id" # Optional
base_url: "https://api.openai.com/v1" # Optional
temperature: 0.7 # Optional, 0.0–2.0
max_tokens: 4096 # Optional
anthropic:
enabled: false
api_key: "your-anthropic-api-key"
model: "claude-sonnet-4-6"
base_url: "https://api.anthropic.com" # Optional
temperature: 0.7 # Optional, 0.0–1.0
max_tokens: 4096 # OptionalYou can enable multiple providers and switch between them at runtime. See docs/configuration.md for full details.
- temperature: Controls the randomness of responses
- Ollama/Gemini/OpenAI/Azure OpenAI: 0.0 (deterministic) to 2.0 (very creative)
- Anthropic: 0.0 (deterministic) to 1.0 (creative)
- max_tokens: Maximum number of tokens in the response (Ollama has no explicit limit)
| Script | Purpose |
|---|---|
bun start |
Run the agent |
bun dev |
Watch mode with auto-reload |
bun build |
Compile to standalone binary devopsagent |
bun test |
Run tests with coverage |
bun run typecheck |
Type check without emitting |
bun run check |
Biome linter and formatter checks |
bun run check:fix |
Auto-fix linting and formatting issues |
bun run format |
Format code with Biome |
bun run format:write |
Format and write changes |
bun run lint |
Run Biome linter |
- API Keys: Store API keys in
config.yaml— it is excluded from version control by default - Command Execution: The execute command tool runs with the same permissions as the Bun process
- Input Validation: All tool parameters are validated against their Zod schemas
This project is licensed under the AGPL-3.0-only License. See the LICENSE file for details.
- Fork the repository
- Create a feature branch
- Make your changes
- Run tests and linting:
bun run check - Submit a pull request
"Provider configuration is missing"
- Ensure
config.yamlexists and contains valid provider configuration - Check that
default_providermatches a key underprovidersthat hasenabled: true
"API key is required"
- Verify your API key in
config.yaml - Make sure the API key has access to the provider's API
Ollama connection refused
- Make sure the Ollama daemon is running:
ollama serve - Verify the model is pulled:
ollama list - Check that
base_urlin config matches where Ollama is listening
Tool execution failures
- Check that the system has the necessary permissions for command execution
- Verify that required system tools are installed
"Must be run in an interactive terminal"
- The Ink TUI requires a TTY. Run the agent directly in a terminal, not piped or redirected.
Markdown rendering issues
- If terminal output appears garbled, ensure your terminal supports ANSI escape codes
- For best results, use a modern terminal emulator (VS Code integrated terminal, iTerm2, Windows Terminal)