/**
* Consolidated configuration
* All environment variables, constants, and LLM config in one place
*/
import { VERSION, PACKAGE_NAME, PACKAGE_DESCRIPTION } from '../version.js';
// Import version utilities (not re-exported - use directly from version.ts if needed externally)
// ============================================================================
// Safe Integer Parsing Helper
// ============================================================================
/**
* Safely parse an integer from environment variable with bounds checking
* @param value - The string value to parse (from process.env)
* @param defaultVal - Default value if parsing fails or value is undefined
* @param min - Minimum allowed value (clamped if below)
* @param max - Maximum allowed value (clamped if above)
* @returns Parsed integer within bounds, or default value
*/
function safeParseInt(
value: string | undefined,
defaultVal: number,
min: number,
max: number
): number {
if (!value) {
return defaultVal;
}
const parsed = parseInt(value, 10);
if (isNaN(parsed)) {
console.warn(`[Config] Invalid number "${value}", using default ${defaultVal}`);
return defaultVal;
}
if (parsed < min) {
console.warn(`[Config] Value ${parsed} below minimum ${min}, clamping to ${min}`);
return min;
}
if (parsed > max) {
console.warn(`[Config] Value ${parsed} above maximum ${max}, clamping to ${max}`);
return max;
}
return parsed;
}
// ============================================================================
// Environment Parsing
// ============================================================================
interface EnvConfig {
SCRAPER_API_KEY: string;
SEARCH_API_KEY: string | undefined;
REDDIT_CLIENT_ID: string | undefined;
REDDIT_CLIENT_SECRET: string | undefined;
}
export function parseEnv(): EnvConfig {
return {
SCRAPER_API_KEY: process.env.SCRAPEDO_API_KEY || '',
SEARCH_API_KEY: process.env.SERPER_API_KEY || undefined,
REDDIT_CLIENT_ID: process.env.REDDIT_CLIENT_ID || undefined,
REDDIT_CLIENT_SECRET: process.env.REDDIT_CLIENT_SECRET || undefined,
};
}
// ============================================================================
// Research API Configuration
// ============================================================================
export const RESEARCH = {
BASE_URL: process.env.OPENROUTER_BASE_URL || 'https://openrouter.ai/api/v1',
MODEL: process.env.RESEARCH_MODEL || 'x-ai/grok-4-fast',
FALLBACK_MODEL: process.env.RESEARCH_FALLBACK_MODEL || 'google/gemini-2.5-flash',
API_KEY: process.env.OPENROUTER_API_KEY || '',
// Timeout: min 1s, max 1hr, default 30min
TIMEOUT_MS: safeParseInt(process.env.API_TIMEOUT_MS, 1800000, 1000, 3600000),
REASONING_EFFORT: (process.env.DEFAULT_REASONING_EFFORT as 'low' | 'medium' | 'high') || 'high',
// Max URLs in search results: min 10, max 200, default 100
MAX_URLS: safeParseInt(process.env.DEFAULT_MAX_URLS, 100, 10, 200),
} as const;
// ============================================================================
// MCP Server Configuration
// ============================================================================
// Version is now automatically read from package.json via version.ts
// No need to manually update version strings anymore!
export const SERVER = {
NAME: PACKAGE_NAME,
VERSION: VERSION,
DESCRIPTION: PACKAGE_DESCRIPTION,
} as const;
// ============================================================================
// Capability Detection (which features are available based on ENV)
// ============================================================================
interface Capabilities {
reddit: boolean; // REDDIT_CLIENT_ID + REDDIT_CLIENT_SECRET
search: boolean; // SERPER_API_KEY
scraping: boolean; // SCRAPEDO_API_KEY
deepResearch: boolean; // OPENROUTER_API_KEY
llmExtraction: boolean; // OPENROUTER_API_KEY (for what_to_extract in scraping)
}
export function getCapabilities(): Capabilities {
const env = parseEnv();
return {
reddit: !!(env.REDDIT_CLIENT_ID && env.REDDIT_CLIENT_SECRET),
search: !!env.SEARCH_API_KEY,
scraping: !!env.SCRAPER_API_KEY,
deepResearch: !!RESEARCH.API_KEY,
llmExtraction: !!RESEARCH.API_KEY, // Reuses OPENROUTER for LLM extraction
};
}
export function getMissingEnvMessage(capability: keyof Capabilities): string {
const messages: Record<keyof Capabilities, string> = {
reddit: 'ā **Reddit tools unavailable.** Set `REDDIT_CLIENT_ID` and `REDDIT_CLIENT_SECRET` to enable.\n\nš Create a Reddit app at: https://www.reddit.com/prefs/apps (select "script" type)',
search: 'ā **Search unavailable.** Set `SERPER_API_KEY` to enable web search and Reddit search.\n\nš Get your free API key at: https://serper.dev (2,500 free queries)',
scraping: 'ā **Web scraping unavailable.** Set `SCRAPEDO_API_KEY` to enable URL content extraction.\n\nš Sign up at: https://scrape.do (1,000 free credits)',
deepResearch: 'ā **Deep research unavailable.** Set `OPENROUTER_API_KEY` to enable AI-powered research.\n\nš Get your API key at: https://openrouter.ai/keys',
llmExtraction: 'ā ļø **AI extraction disabled.** The `use_llm` and `what_to_extract` features require `OPENROUTER_API_KEY`.\n\nScraping will work but without intelligent content filtering.',
};
return messages[capability];
}
// ============================================================================
// Scraper Configuration (Scrape.do implementation)
// ============================================================================
export const SCRAPER = {
MAX_CONCURRENT: 30,
BATCH_SIZE: 30,
MAX_TOKENS_BUDGET: 32000,
MIN_URLS: 3,
MAX_URLS: 50,
RETRY_COUNT: 3,
RETRY_DELAYS: [2000, 4000, 8000] as const,
EXTRACTION_SUFFIX: 'Try to answer this information as comprehensive as possible while keeping info density super high without adding unnecessary words but satisfy the scope defined by previous instructions even more.',
} as const;
// ============================================================================
// Reddit Configuration
// ============================================================================
export const REDDIT = {
MAX_CONCURRENT: 10,
BATCH_SIZE: 10,
MAX_COMMENT_BUDGET: 1000,
MAX_COMMENTS_PER_POST: 200,
MIN_POSTS: 2,
MAX_POSTS: 50,
RETRY_COUNT: 5,
RETRY_DELAYS: [2000, 4000, 8000, 16000, 32000] as const,
} as const;
// ============================================================================
// CTR Weights for URL Ranking (inspired from CTR research)
// ============================================================================
export const CTR_WEIGHTS: Record<number, number> = {
1: 100.00,
2: 60.00,
3: 48.89,
4: 33.33,
5: 28.89,
6: 26.44,
7: 24.44,
8: 17.78,
9: 13.33,
10: 12.56,
} as const;
// ============================================================================
// LLM Extraction Model (uses OPENROUTER for scrape_links AI extraction)
// ============================================================================
export const LLM_EXTRACTION = {
MODEL: process.env.LLM_EXTRACTION_MODEL || 'openai/gpt-oss-120b:nitro',
MAX_TOKENS: 8000,
ENABLE_REASONING: process.env.LLM_ENABLE_REASONING !== 'false', // Default true, can be disabled with 'false'
} as const;
