import type { LTApiResult, LTApiAuth } from '../types/sdk';
/**
 * Execute a natural-language query against connected MCP servers.
 *
 * Requires an LLM API key (OPENAI_API_KEY or ANTHROPIC_API_KEY) to be
 * configured. Returns 503 if neither is set. Timeouts surface as 504.
 *
 * @param input.prompt — the natural-language query to execute (required)
 * @param input.tags — optional tags to scope which MCP servers are queried
 * @param input.wait — when true, blocks until the query completes; otherwise returns immediately
 * @param input.direct — when true, bypasses workflow orchestration and queries the LLM directly
 * @param input.context — optional additional context forwarded to the LLM
 * @param auth — authenticated user context
 * @returns `{ status: 200, data: { ... } }` query result from the MCP pipeline
 */
export declare function mcpQuery(input: {
    prompt: string;
    tags?: string[];
    wait?: boolean;
    direct?: boolean;
    context?: any;
}, auth?: LTApiAuth): Promise<LTApiResult>;
/**
 * Generate a workflow definition from a natural-language description.
 *
 * Uses an LLM to produce a YAML workflow from the given prompt. Supports
 * iterative refinement by accepting prior YAML, feedback, and Q&A answers.
 * Requires an LLM API key. Timeouts surface as 504.
 *
 * @param input.prompt — natural-language description of the desired workflow (required)
 * @param input.tags — optional tags to scope available MCP tools
 * @param input.wait — when true, blocks until generation completes
 * @param input.feedback — optional refinement feedback on a previous generation
 * @param input.prior_yaml — optional YAML from a previous generation to refine
 * @param input.answers — optional answers to clarifying questions from a prior round
 * @param input.prior_questions — optional questions from a prior round for context
 * @param auth — authenticated user context
 * @returns `{ status: 200, data: { ... } }` generated workflow definition
 */
export declare function buildWorkflow(input: {
    prompt: string;
    tags?: string[];
    wait?: boolean;
    feedback?: string;
    prior_yaml?: string;
    answers?: any;
    prior_questions?: any;
}, auth?: LTApiAuth): Promise<LTApiResult>;
/**
 * Refine an existing workflow definition using feedback.
 *
 * Takes a previously generated YAML workflow, the original prompt, and
 * user feedback, then produces an updated workflow definition.
 *
 * @param input.prompt — original natural-language description (required)
 * @param input.prior_yaml — the YAML workflow to refine (required)
 * @param input.feedback — user feedback describing desired changes (required)
 * @param input.tags — optional tags to scope available MCP tools
 * @param input.wait — when true, blocks until refinement completes
 * @param auth — authenticated user context
 * @returns `{ status: 200, data: { ... } }` refined workflow definition
 */
export declare function refineWorkflow(input: {
    prompt: string;
    prior_yaml: string;
    feedback: string;
    tags?: string[];
    wait?: boolean;
}, auth?: LTApiAuth): Promise<LTApiResult>;
/**
 * Generate a human-readable description and tags for a workflow.
 *
 * Uses an LLM to produce a concise description from the prompt and
 * optional result metadata. Falls back to the raw prompt as the
 * description if the LLM call fails.
 *
 * @param input.prompt — the workflow prompt or content to describe (required)
 * @param input.result_title — optional title from the workflow result for additional context
 * @param input.result_summary — optional summary from the workflow result for additional context
 * @returns `{ status: 200, data: { description, tags } }` generated description and tag array
 */
export declare function describeWorkflow(input: {
    prompt: string;
    result_title?: string;
    result_summary?: string;
}): Promise<LTApiResult>;