engr

Claude Code wrapper with provider switching. Run Claude Code against Anthropic (default), MiniMax, or Ollama without manual config changes.

Install

./install

Installs to ~/.local/bin by default. Pass a different path:

./install /usr/local/bin

If the install directory isn't in your PATH, the script will print the line to add to your shell config.

Usage

engr                        # Anthropic (default)
engr --minimax              # MiniMax provider
engr --ollama               # Ollama provider
engr --ollama -m glm-5:cloud  # Ollama with model override
engr --glm                    # GLM coding plan (z.ai)
engr --kimi                   # Kimi provider (Moonshot)
engr --openrouter             # OpenRouter provider
engr --openai                 # OpenAI provider
engr --litellm                # LiteLLM proxy
engr --bifrost                # MaximHQ Bifrost gateway
engr -y                       # Skip permission prompts (yolo mode)
engr [any claude args]        # All other args pass through to claude

Providers

Anthropic (default)

No extra setup. Uses whatever Anthropic credentials are already configured.

MiniMax

Requires an API key:

export ENGR_MINIMAX_API_KEY="your-key"
engr --minimax

Ollama

Requires Ollama installed. The script auto-starts the Ollama server and pulls the model if not found locally.

engr --ollama                # Default model
engr --ollama -m glm-5:cloud # Specific model

GLM coding plan (z.ai)

Requires a z.ai API key:

export ENGR_ZAI_API_KEY="your-key"
engr --glm

Kimi (Moonshot)

Requires a Moonshot API key:

export ENGR_KIMI_API_KEY="your-key"
engr --kimi

Uses the kimi-k2.5 model on api.moonshot.ai.

OpenRouter

Requires an API key:

export ENGR_OPENROUTER_API_KEY="your-key"
engr --openrouter

Uses the OpenRouter SDK compatibility layer to route calls through the Claude API at https://api.anthropic.com/v1/.

OpenAI

Requires an API key:

export ENGR_OPENAI_API_KEY="your-key"
engr --openai

Uses the OpenAI SDK compatibility layer to route calls through the Claude API at https://api.anthropic.com/v1/.

LiteLLM

Requires LiteLLM proxy running. Set your API key and optionally the proxy URL:

export ENGR_LITELLM_API_KEY="your-master-key"
engr --litellm

Defaults to http://localhost:4000. Override with ENGR_LITELLM_BASE_URL:

export ENGR_LITELLM_BASE_URL="https://litellm.mycompany.com"
engr --litellm

Pass a model override:

engr --litellm -m gpt-4o

MaximHQ Bifrost

Requires a Maxim workspace API key and Bifrost gateway URL:

export ENGR_BIFROST_API_KEY="maxim-xxxx"
export ENGR_BIFROST_BASE_URL="https://<your-workspace>.bifrost.getmaxim.ai"
engr --bifrost

Pass a model override:

engr --bifrost -m anthropic/claude-sonnet-4-5

Yolo mode

By default, engr runs Claude Code with permission prompts enabled. Use -y or --yolo to skip all permission prompts:

engr -y                      # Skip permission prompts
engr --yolo --openrouter     # OpenRouter + skip permissions

This passes --dangerously-skip-permissions to Claude Code. Use with caution.

Help

engr --help