/**
 * Core Axiom Block Builder — shared prompt section for CORE_PRINCIPLES injection.
 *
 * Provides a single, consistent way to inject the core principle list
 * into any prompt builder. Used by the OpenClaw prompt hook (foundational
 * axioms → `<core_principles>`), the diagnostician pipeline (rootcause,
 * distiller), and the internalization pipeline (dreamer, philosopher, scribe).
 *
 * ## Design
 *
 * - `formatCorePrinciplesList()` — low-level: just the formatted list lines
 * - `buildCoreAxiomBlock()` — high-level: conditional block with header + instruction
 *
 * Both support bilingual output via `outputLanguage` and layer scoping via
 * `scope` (default `'all'`: the full 10-principle set for LLM grounding).
 */
import type { OutputLanguage } from '../language-directive.js';
/**
 * Which registry slice a formatter emits. The registry is the only
 * classification source — callers never hand-write id filters.
 * - 'foundational' — foundational axioms (the `<core_principles>` set)
 * - 'operating'    — operating principles
 * - 'all'          — the full 10-principle set; default, used for LLM axiom
 *                    grounding in diagnostician/internalization prompts
 */
export type CorePrincipleScope = 'foundational' | 'operating' | 'all';
export interface CoreAxiomBlockOptions {
    /** Whether to include the core axioms section (default: false). */
    coreGrounding?: boolean;
    /**
     * Section title. Defaults to 'CORE AXIOMS:'.
     * Override for context-specific headers (e.g. 'PHASE 3.5 — Core Axiom Grounding:').
     */
    sectionTitle?: string;
    /**
     * Instruction text after the title, before the list.
     * Defaults to a generic instruction about referencing axiom IDs.
     * Set to '' to omit.
     */
    instruction?: string;
    /** Output language for bilingual principle statements. */
    outputLanguage?: OutputLanguage;
    /** Which registry slice to list (default: 'active'). */
    scope?: CorePrincipleScope;
    /** Fallback string when coreGrounding is false (default: ''). */
    fallback?: string;
}
/**
 * Format a core principle list as a newline-separated string.
 *
 * Each line: `T-XX: <statement>`
 * When `outputLanguage` is 'zh-CN', uses `statementZh` instead of `statement`.
 *
 * @param outputLanguage - Optional language override for bilingual output.
 * @param scope - Which registry slice to list (default: 'all').
 * @returns Formatted list string (e.g. "T-01: Build a sufficient model...\nT-02: ...")
 */
export declare function formatCorePrinciplesList(outputLanguage?: OutputLanguage, scope?: CorePrincipleScope): string;
/**
 * Build a conditional core axiom block for prompt injection.
 *
 * When `coreGrounding` is true, returns a formatted block containing:
 * - Section title
 * - Instruction text
 * - Formatted principle list
 *
 * When `coreGrounding` is false/undefined, returns `fallback` (default: empty string).
 *
 * ## Usage
 *
 * ```typescript
 * // In a prompt builder:
 * const axiomBlock = buildCoreAxiomBlock({ coreGrounding: true, outputLanguage: 'zh-CN' });
 * return `...existing prompt...\n${axiomBlock}\n...rest of prompt...`;
 * ```
 *
 * @param opts - Block options (see CoreAxiomBlockOptions).
 * @returns Formatted block string, or fallback when coreGrounding is off.
 */
export declare function buildCoreAxiomBlock(opts?: CoreAxiomBlockOptions): string;
//# sourceMappingURL=core-axiom-block.d.ts.map