Settings & Diagnostics

KnightCode settings are managed directly inside the terminal workspace using interactive dialog boxes.

Settings Dialogs

Use slash commands to open individual settings dialogs:

• /models: View and switch the active AI model (all models route through OpenRouter).
• /theme: Choose your terminal colors and styling themes.
• /reasoning: Set the thinking/reasoning effort level (None, Low, Medium, High, Max) for models that support it. This adjusts the thinking token budget.
• /allow: Manage allowed command prefixes. A Bash command that matches a listed prefix runs automatically without prompting for confirmation.
• /hooks: Manage lifecycle hook scripts (PreToolUse, PostToolUse, PostToolUseFailure, UserPromptSubmit, Stop) to enforce constraints or trigger linters/formatters.

Diagnostics (The Doctor Command)

If you encounter issues with API key authorization, database locks, or environment configurations, you can run a diagnostic audit:

In-App Diagnostics

Type /doctor inside the TUI session. A dialog opens and runs four checks:

• OpenRouter API key: Verifies an API key is available (from credentials.json or the environment). A missing key is a warning, not a failure.
• Local store: Confirms the local SQLite store is readable and ready.
• Git available: Checks that git is installed and reports its version.
• Runtime: Reports the runtime version KnightCode is running on.

CLI Diagnostics

If the TUI is failing to launch, run the diagnostic tool in headless mode from your standard terminal shell:

knightcode doctor

This performs the same audit checks and prints a clean diagnostics summary directly to stdout, exiting with a non-zero code if critical failures are found.