Skip to content
Documentation sections

CLI configuration

Use a runtime-loadable JavaScript or JSON file. TypeScript config files are intentionally rejected with a warning.

krythiq.config.mjs
import { defineConfig } from "krythiq"; export default defineConfig({  model: "claude-sonnet-4-20250514",  traces: { enabled: true },  scan: {    ignore: ["generated/**"],    ai: false,    aiModel: "mistral-large-latest",    publish: false,    rules: ".krythiq/rules.json"  }});
Engine and output controls

Use scan --no-semgrep or --no-audit to skip an external engine. These do not disable AI summaries, saving, or publishing enabled separately. There is no built-in rule pack; reviewed custom rules are needed when both engines are off.

Use --format json --output .krythiq/reports/scan.json to retain evidence. Paths and output formats are explicit CLI options. Existing files are preserved unless --overwrite is passed.

doctor [path] loads the configuration in that directory and reports local prerequisites. Config files are trusted executable JavaScript.

Recognized files and precedence

The first existing file is loaded: krythiq.config.mjs, .js, .cjs, .json, then .krythiq/config.json.

CLI options select rules, AI model, and publishing when provided. Environment flags can also enable AI summaries or publishing. Trace model precedence is the explicit --model, then environment, then config, then the built-in default.