Quickstart

1. Start a model backend

For the local examples:

ollama serve

In another terminal:

ollama pull qwen3.5:latest

Alternative endpoints:

Backend	Setup	Endpoint
llama.cpp	llama-server -m model.gguf --port 8000	http://localhost:8000/v1
LM Studio	Load a model and start its server	http://localhost:1234/v1
vLLM	Start a compatible model server	Its /v1 base URL

2. Configure the model

~/.gdc/config.toml

[provider]
base_url = "http://localhost:11434/v1"
model = "qwen3.5:latest"

Check connectivity:

gdc chat "say hello"

3. Launch the interactive app

cd ~/my-project
gdc-tui

Type a request and press Enter:

list the source files and describe the project

For approval prompts, a allows, r toggles remembering and d denies. Press Ctrl+C twice to exit. See Terminal.

4. Run a headless task

gdc print "list the Rust files in this project"

For unattended requests that require approval:

gdc print --auto-approve "run the tests and fix failures"

See Headless CLI.

5. Review what happened

gdc sessions list
gdc sessions show <id>

For history created by the local TUI:

gdc sessions list --all-projects
gdc --db-path ~/.gdc/projects/<slug>/sessions.db sessions show <id>

Overriding settings for one run

gdc --base-url http://localhost:8000/v1 --model loaded-model \
  print "explain error handling in src/app.rs"

Next: Permissions, Plan mode, Configuration.