This page gets a fresh checkout to the first audited, operator-approved LinuxAgent turn.
git clone https://github.com/Eilen6316/LinuxAgent.git
cd LinuxAgent
./scripts/bootstrap.shThe bootstrap script installs project dependencies in .venv, creates a
user-level ~/.local/bin/linuxagent launcher, and writes
LINUXAGENT_CONFIG=$HOME/.config/linuxagent/config.yaml to your shell profile.
Open a new shell or run source ~/.bashrc before starting linuxagent from
another directory. If the command is not found, add ~/.local/bin to PATH.
Edit ~/.config/linuxagent/config.yaml and configure one provider.
Remote provider:
api:
provider: deepseek
api_key: "your-real-key"Local OpenAI-compatible provider:
api:
provider: ollama
base_url: http://127.0.0.1:11434/v1
model: llama3.1
api_key: ""
token_parameter: max_tokensKeep the config file owned by your user and private:
chmod 600 ~/.config/linuxagent/config.yamllinuxagent checkFix any reported config or provider issue before continuing.
linuxagentTry a read-only first request:
check the Linux version
When LinuxAgent proposes the first LLM-generated command, the confirmation menu
lets you run it once, allow the same argv command shape for this conversation
and the same /resume thread, or refuse it.
Use /resume to reopen a saved conversation. Use /new to start a fresh
conversation inside the running CLI.