/**
 * Spider Skill - Fast web scraping and crawling capabilities.
 *
 * Tier 2 built-in skill. Port of the Python `SpiderSkill` that performs
 * HTTP scraping via `fetch`, lightweight HTML-to-text extraction, and
 * breadth-first crawling with configurable limits. Supports three tools:
 *
 * - `scrape_url` — single-page text/markdown extraction
 * - `crawl_site` — breadth-first crawl from a start URL
 * - `extract_structured_data` — CSS/XPath-like structured extraction
 *
 * Multiple instances are supported; use `tool_name` in config to differentiate.
 */
import { SkillBase } from '../SkillBase.js';
import type { SkillToolDefinition, SkillConfig, ParameterSchemaEntry } from '../SkillBase.js';
/**
 * Fast web scraping skill optimized for speed and token efficiency.
 *
 * Multi-instance capable. Port of the Python `SpiderSkill` with three tools:
 * `scrape_url`, `crawl_site`, and `extract_structured_data`. Configuration
 * mirrors the Python schema (delay, concurrent_requests, timeout, max_pages,
 * max_depth, extract_type, max_text_length, clean_text, selectors,
 * follow_patterns, user_agent, headers, follow_robots_txt, cache_enabled).
 *
 * @example
 * ```ts
 * agent.addSkill('spider', { max_pages: 5, max_depth: 2 });
 * ```
 */
export declare class SpiderSkill extends SkillBase {
    static SKILL_NAME: string;
    static SKILL_DESCRIPTION: string;
    static SKILL_VERSION: string;
    static REQUIRED_PACKAGES: readonly string[];
    static REQUIRED_ENV_VARS: readonly string[];
    static SUPPORTS_MULTIPLE_INSTANCES: boolean;
    static getParameterSchema(): Record<string, ParameterSchemaEntry>;
    private delay;
    private concurrentRequests;
    private timeout;
    private maxPages;
    private maxDepth;
    private extractType;
    private maxTextLength;
    private cleanText;
    private cacheEnabled;
    private followRobotsTxt;
    private userAgent;
    private headers;
    private selectors;
    private compiledFollowPatterns;
    private cache;
    private readonly cacheMaxSize;
    getInstanceKey(): string;
    setup(): Promise<boolean>;
    getHints(): string[];
    cleanup(): Promise<void>;
    /** @returns Three tools: `scrape_url`, `crawl_site`, and `extract_structured_data`. */
    getTools(): SkillToolDefinition[];
    /**
     * Redirect a target URL through `SPIDER_BASE_URL` when set. Used by the
     * porting-sdk's `audit_skills_dispatch.py` to point every outbound fetch
     * at its loopback fixture without requiring the test URL to resolve.
     * Preserves the path (and any query/fragment) from the original target.
     */
    private static _redirectForAudit;
    /** Fetch a URL with caching and timeout handling. Returns null on failure. */
    private _fetchUrl;
    /**
     * Extract plain text from a fetched response using simple regex stripping.
     *
     * Takes a {@link CachedResponse} (TS equivalent of `requests.Response`) to
     * match Python's `_fast_text_extract(response)` capability surface — the
     * response-like object exposes `url`, `status`, and `body` for callers that
     * need to branch on status/content-type.
     */
    private _fastTextExtract;
    /**
     * Extract a value using a CSS selector (via cheerio) or XPath-like
     * selector (`//tag`). Mirrors Python `_structured_extract`'s selector
     * branching: selectors starting with `/` are treated as XPath (tag-name
     * subset), everything else is full CSS via cheerio's querySelector-style
     * engine.
     */
    private _applySelector;
    /** Markdown extraction using cheerio. Mirrors Python `_markdown_extract`. */
    private _markdownExtract;
    private _structuredExtract;
    private _scrapeUrlHandler;
    private _crawlSiteHandler;
    private _extractStructuredHandler;
}
/**
 * Factory function for creating SpiderSkill instances.
 * @param config - Optional skill configuration.
 * @returns A new SpiderSkill instance.
 */
export declare function createSkill(config?: SkillConfig): SpiderSkill;