Prerequisites
Git Required to clone the repository and for Go to download dependencies.
Go 1.26+ Only needed for go install or building from source.
API Key or OAuth An API key from an LLM provider or an account with an active plan for OAuth.
1. Installation Homebrew (Recommended)
Binary
go install
From Source
Docker
The easiest way for macOS and Linux . Installs and updates automatically. brew tap diillson/chatcli brew install chatcli
To upgrade: Supports macOS (Apple Silicon and Intel) and Linux (amd64). The formula is automatically updated on each release.
Download the binary for your system directly from the Releases page . # Linux/macOS — adjust the filename according to the release chmod +x chatcli sudo mv chatcli /usr/local/bin/
Available for Linux , macOS , and Windows on amd64 and arm64 architectures.
If you have Go installed: go install github.com/diillson/chatcli@latest
The binary will be installed in $GOPATH/bin. Make sure it is in your PATH: export PATH = $PATH :$( go env GOPATH )/ bin
Add the line above to your .bashrc or .zshrc to persist across sessions. For those who want to compile locally or contribute:
Clone the repository
git clone https://github.com/diillson/chatcli.git cd chatcli
Build
go mod tidy go build -o chatcli
(Optional) Build with version information
VERSION = $( git describe --tags --always --dirty 2> /dev/null || echo "dev" ) COMMIT_HASH = $( git rev-parse --short HEAD ) BUILD_DATE = $( date -u +"%Y-%m-%dT%H:%M:%SZ" ) go build -ldflags " \ -X github.com/diillson/chatcli/version.Version=${ VERSION } \ -X github.com/diillson/chatcli/version.CommitHash=${ COMMIT_HASH } \ -X github.com/diillson/chatcli/version.BuildDate=${ BUILD_DATE }" \ -o chatcli main.go
This injects version data into the binary, accessible via /version or chatcli --version.
Move to PATH
sudo mv chatcli /usr/local/bin/
Create a .env file in your home directory or in the project root.
OpenAI
Anthropic (Claude)
Google (Gemini)
xAI (Grok)
StackSpot AI
ZAI (Zhipu AI)
MiniMax
Moonshot (Kimi)
OpenRouter
OAuth (no API key)
Ollama (Local)
LLM_PROVIDER=OPENAI OPENAI_API_KEY="sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx" # (Optional) Specific model — default: gpt-5.4 # OPENAI_MODEL="gpt-5.4"
LLM_PROVIDER=CLAUDEAI ANTHROPIC_API_KEY="sk-ant-xxxxxxxxxxxxxxxxxxxxxxxx" # (Optional) Model — default: claude-sonnet-4-6 # ANTHROPIC_MODEL="claude-sonnet-4-6"
LLM_PROVIDER=GOOGLEAI GOOGLEAI_API_KEY="AIzaxxxxxxxxxxxxxxxxxxxxxxxx" # (Optional) Model — default: gemini-2.5-flash # GOOGLEAI_MODEL="gemini-2.5-pro"
LLM_PROVIDER=XAI XAI_API_KEY="xai-xxxxxxxxxxxxxxxxxxxxxxxx" # (Optional) Model — default: grok-4-1 # XAI_MODEL="grok-4-1"
LLM_PROVIDER=STACKSPOT CLIENT_ID="your-client-id" CLIENT_KEY="your-client-secret" # (Optional) Realm/tenant — default: zup # STACKSPOT_REALM="your-tenant-name" # (Optional) Agent ID — default: default # STACKSPOT_AGENT_ID="your-agent-id"
Authentication uses OAuth 2.0 Client Credentials . Get your credentials from the StackSpot portal.
You can also override realm and agent-id via command-line flags: chatcli --provider STACKSPOT --realm "my-realm" --agent-id "my-agent" -p "Your prompt"
LLM_PROVIDER=ZAI ZAI_API_KEY="your-zai-api-key" # (Optional) Model — default: glm-5 # ZAI_MODEL="glm-4.7"
LLM_PROVIDER=MINIMAX MINIMAX_API_KEY="your-minimax-api-key" # (Optional) Model — default: MiniMax-M2.7 (case-sensitive!) # MINIMAX_MODEL="MiniMax-M2.5"
LLM_PROVIDER=MOONSHOT MOONSHOT_API_KEY="sk-your-moonshot-api-key" # (Optional) Model — default: kimi-k2.6 # MOONSHOT_MODEL="kimi-k2.6" # MOONSHOT_THINKING="auto" # auto | enabled | disabled
LLM_PROVIDER=OPENROUTER OPENROUTER_API_KEY="sk-or-xxxxxxxxxxxxxxxxxxxxxxxx" # (Optional) Model — default: openai/gpt-4o # MODEL="anthropic/claude-sonnet-4" # (Optional) Server-side fallback routing # OPENROUTER_FALLBACK_MODELS="anthropic/claude-sonnet-4,google/gemini-2.5-flash"
OpenRouter is a multi-provider gateway — one API key gives access to 200+ models from OpenAI, Anthropic, Google, Meta, Mistral, DeepSeek, and more. Get your key at openrouter.ai . If you have ChatGPT Plus/Codex , Claude Pro , or GitHub Copilot : chatcli # Inside interactive mode: /auth login openai-codex # OpenAI (PKCE OAuth) /auth login anthropic # Anthropic (PKCE OAuth) /auth login github-copilot # GitHub Copilot (Device Flow)
The browser will open automatically. See the OAuth documentation for details. LLM_PROVIDER=OLLAMA OLLAMA_ENABLED=true OLLAMA_BASE_URL="http://localhost:11434" OLLAMA_MODEL="llama3"
No API key required. Install Ollama from ollama.com and download a model with ollama pull llama3. You only need to configure the providers you plan to use. ChatCLI automatically detects which ones are available based on the keys found.
Additional Environment Variables Variable Description Default CHATCLI_DOTENVCustom path for the .env file .envCHATCLI_LANGForce a specific language (e.g., pt-BR, en) Auto-detected LOG_LEVELLog level: debug, info, warn, error infoCHATCLI_ENVLogging mode: dev (colored console + file), prod (file-only JSON) prodMAX_RETRIESMaximum retries for API calls 5
For the complete list of variables, see the Environment Variables Reference .
3. Verify the Installation
Check the active configuration
chatcli # Inside interactive mode: /config
Ask your first question
What is the Dirac equation?
If you get a response from the AI, congratulations! ChatCLI is ready to use.
Next Steps
Basic Usage Essential commands, modes, and navigation in the interactive prompt.
Agent Mode Delegate tasks for the AI to execute in your terminal.
Coder Mode AI that reads, edits, and tests code in a loop.
Skill Registry Search and install skills from remote registries.
