Configuration

DojOps supports 6 LLM providers with flexible configuration via CLI flags, environment variables, config files, and named profiles.

Supported Providers

Provider	`DOJOPS_PROVIDER`	Required Env Var	Default Model	SDK
OpenAI	`openai`	`OPENAI_API_KEY`	`gpt-4o-mini`	`openai`
Anthropic	`anthropic`	`ANTHROPIC_API_KEY`	`claude-sonnet-4-5-20250929`	`@anthropic-ai/sdk`
Ollama	`ollama`	(none — local)	`llama3`	`ollama`
DeepSeek	`deepseek`	`DEEPSEEK_API_KEY`	`deepseek-chat`	`openai` (compatible)
Gemini	`gemini`	`GEMINI_API_KEY`	`gemini-2.5-flash`	`@google/genai`
GitHub Copilot	`github-copilot`	(OAuth Device Flow)	`gpt-4o`	`openai` (compatible)

Provider Management

DojOps includes a dedicated provider command for managing LLM providers — adding, removing, switching, and listing them.

Adding Providers


# Add your first provider (auto-set as default)
dojops provider add openai --token sk-...
 
# Add a second provider (preserves existing default)
dojops provider add anthropic --token sk-ant-...
 
# Ollama (local, no token needed)
dojops provider add ollama

Switching Providers


# Interactive picker — shows only configured providers
dojops provider switch
 
# Direct — set by name
dojops provider default anthropic
 
# Shortcut flag
dojops provider --as-default openai

Listing Providers


# Table view (default)
dojops provider
 
# JSON output
dojops provider list --output json

Output shows configured (*) vs unconfigured (o) providers, with the default marked:


┌ Providers ──────────────────────────────────────┐
│  * anthropic         sk-***ntx                  │
│  * openai (default)  sk-***ojx  model: gpt-4o   │
│  o deepseek          (not set)                  │
│  o gemini            (not set)                  │
│  * ollama            (local)                    │
└─────────────────────────────────────────────────┘

Removing Providers


dojops provider remove openai

If the removed provider was the default, DojOps clears the default and suggests an alternative.

Configuration Methods

Interactive Setup


dojops config

The interactive wizard:

Selects a provider from the list
Prompts for the API key (skipped for Ollama and GitHub Copilot)
For GitHub Copilot: runs the OAuth Device Flow automatically if not already authenticated
Fetches available models from the provider’s API via listModels()
Shows an interactive model picker
Asks whether to switch default provider (if one already exists)
Offers to configure another provider

Environment Variables


export DOJOPS_PROVIDER=openai
export OPENAI_API_KEY=sk-...
export DOJOPS_MODEL=gpt-4o              # Optional model override
export DOJOPS_TEMPERATURE=0.7           # Optional temperature override
export DOJOPS_API_PORT=3000              # API server port (default: 3000)

CLI Flags


dojops --provider=anthropic "Create a Terraform config"
dojops --model=gpt-4o "Create a Kubernetes deployment"
dojops --temperature=0.2 "Create a Terraform config"

Config File

DojOps saves configuration to ~/.dojops/config.json:


{
  "provider": "openai",
  "model": "gpt-4o-mini",
  "defaultTemperature": 0.7,
  "token": "sk-..."
}

Configuration Precedence

Values are resolved in order (first match wins):


Provider:     --provider flag     >  $DOJOPS_PROVIDER     >  config file  >  "openai" (default)
Model:        --model flag        >  $DOJOPS_MODEL        >  config file  >  provider default
Temperature:  --temperature flag  >  $DOJOPS_TEMPERATURE  >  config file  >  undefined (provider default)
              Note: `apply --replay` forces temperature=0 regardless of other settings
Token:        $OPENAI_API_KEY (etc.)  >  config file token

Model Selection

Each provider ships with a sensible default, but you can choose any model your provider supports:


# Interactive: fetches models from provider API, shows picker
dojops config
 
# Set directly in config
dojops config --model=gpt-4o
 
# One-off override (doesn't change saved config)
dojops --model=deepseek-reasoner "Analyze this Terraform plan"

Dynamic Model Discovery

When running dojops config, DojOps calls the provider’s listModels() API to fetch available models:

OpenAI — Lists models from the OpenAI API
Anthropic — Lists supported Claude models
Ollama — Lists locally installed models
DeepSeek — Lists available DeepSeek models
Gemini — Lists available Gemini models
GitHub Copilot — Lists models available to your Copilot subscription tier

Profiles

Named profiles let you switch between different provider/environment configurations:

Create a Profile


# Save current config as a named profile
dojops config profile create staging

Use a Profile


# Switch to a profile (updates active config)
dojops config profile use staging
 
# One-off profile override (doesn't change active config)
dojops --profile=staging "Create an S3 bucket"

List Profiles


dojops config profile list

Example: Multi-Environment Setup


# Set up development profile (local Ollama)
export DOJOPS_PROVIDER=ollama
dojops config
dojops config profile create dev
 
# Set up production profile (OpenAI)
export DOJOPS_PROVIDER=openai
export OPENAI_API_KEY=sk-prod-...
dojops config --model=gpt-4o
dojops config profile create prod
 
# Switch between them
dojops config profile use dev    # Uses local Ollama
dojops config profile use prod   # Uses OpenAI GPT-4o

Environment Variables Reference

Variable	Description	Default
`DOJOPS_PROVIDER`	LLM provider name	`openai`
`DOJOPS_MODEL`	Model override	Provider default
`DOJOPS_TEMPERATURE`	Temperature override	Provider default
`OPENAI_API_KEY`	OpenAI API key	—
`ANTHROPIC_API_KEY`	Anthropic API key	—
`DEEPSEEK_API_KEY`	DeepSeek API key	—
`GEMINI_API_KEY`	Google Gemini API key	—
`DOJOPS_API_PORT`	API server port	`3000`
`OLLAMA_HOST`	Ollama server URL	`http://localhost:11434`
`OLLAMA_TLS_REJECT_UNAUTHORIZED`	TLS cert verification for Ollama	`true`
`GITHUB_COPILOT_TOKEN`	GitHub OAuth token (skip device flow)	—
`DOJOPS_HUB_URL`	DojOps Hub API base URL	`https://hub.dojops.ai`
`DOJOPS_HUB_TOKEN`	API token for hub publishing (generate one)	—

Ollama Setup

Ollama runs locally and doesn’t require an API key:


# Install Ollama
curl -fsSL https://ollama.com/install.sh | sh
 
# Pull a model
ollama pull llama3
 
# Configure DojOps
export DOJOPS_PROVIDER=ollama
dojops "Create a Dockerfile for Node.js"

By default, DojOps connects to Ollama at http://localhost:11434. To use a remote or custom Ollama server:


# Interactive setup (prompts for URL + TLS settings)
dojops config
 
# Or set via environment variable
export OLLAMA_HOST=https://ollama.corp.internal:8443
 
# For self-signed certificates behind proxies
export OLLAMA_TLS_REJECT_UNAUTHORIZED=false

The Ollama host URL is resolved with this priority: OLLAMA_HOST env > ~/.dojops/config.json > http://localhost:11434.

GitHub Copilot Setup

GitHub Copilot uses an OAuth Device Flow instead of a static API key. You need an active GitHub Copilot subscription (Pro, Pro+, Business, or Enterprise).


# Authenticate via Device Flow (opens browser)
dojops auth login --provider github-copilot
 
# Or use the provider command
dojops provider add github-copilot
 
# Or configure interactively
dojops config
# Select "github-copilot" → Device Flow runs automatically → model picker shown

The Device Flow works as follows:

DojOps requests a one-time device code from GitHub
You open https://github.com/login/device in your browser and enter the code
DojOps polls GitHub for authorization
Once approved, DojOps exchanges the OAuth token for a short-lived Copilot JWT
The JWT is cached and auto-refreshed before each API call (~30 min expiry)
DojOps fetches available models and shows an interactive picker to select the default model

Tokens are stored in ~/.dojops/copilot-token.json (mode 0600). The Copilot API is OpenAI-compatible, so you can use models like gpt-4o, gpt-4o-mini, claude-3.5-sonnet, o1-mini, etc. depending on your subscription tier. After authentication, all three commands (auth login, provider add, and config) will present a model picker so you can choose your preferred model immediately.

For CI/CD, you can set GITHUB_COPILOT_TOKEN to a GitHub OAuth token (ghu_xxx) to skip the interactive Device Flow.

Viewing Configuration


# Show current config
dojops config show
 
# Show system health and config
dojops doctor
 
# Inspect detailed config state
dojops inspect config

`.env` File

For development, create a .env file in the project root:


# .env
DOJOPS_PROVIDER=openai
OPENAI_API_KEY=sk-...
DOJOPS_MODEL=gpt-4o-mini
DOJOPS_API_PORT=3000

See .env.example in the repository for a template.