M.I.M.I.R - Multi-agent Intelligent Memory & Insight Repository

/** * Rate Limiter Configuration * * Defines rate limits for different LLM providers. * Set requestsPerHour to -1 to bypass rate limiting entirely. */ export interface RateLimitSettings { requestsPerHour: number; enableDynamicThrottling: boolean; warningThreshold: number; logLevel: 'silent' | 'normal' | 'verbose'; } export const DEFAULT_RATE_LIMITS: Record<string, RateLimitSettings> = { copilot: { requestsPerHour: 2500, // Conservative limit (50% of 5000 to account for estimation errors) enableDynamicThrottling: true, warningThreshold: 0.80, // Warn at 80% capacity logLevel: 'normal', }, ollama: { requestsPerHour: -1, // Bypass rate limiting for local models enableDynamicThrottling: false, warningThreshold: 0.80, logLevel: 'silent', }, openai: { requestsPerHour: 3000, // Default for OpenAI API enableDynamicThrottling: true, warningThreshold: 0.80, logLevel: 'normal', }, anthropic: { requestsPerHour: 1000, // Conservative for Claude API enableDynamicThrottling: true, warningThreshold: 0.80, logLevel: 'normal', }, }; /** * Load rate limit configuration for a specific provider * * Retrieves the default rate limit settings for a provider and applies * any custom overrides. Falls back to copilot settings if provider is unknown. * * Rate limiting helps prevent API quota exhaustion and ensures fair resource * usage across multiple agents or concurrent requests. * * @param provider - LLM provider name ('copilot', 'ollama', 'openai', 'anthropic') * @param overrides - Optional partial settings to override defaults * @returns Complete rate limit configuration with all settings * * @example * ```ts * // Load default copilot settings * const config = loadRateLimitConfig('copilot'); * console.log(config.requestsPerHour); // 2500 * * // Load with custom overrides * const customConfig = loadRateLimitConfig('openai', { * requestsPerHour: 5000, * logLevel: 'verbose' * }); * * // Disable rate limiting for local models * const ollamaConfig = loadRateLimitConfig('ollama'); * console.log(ollamaConfig.requestsPerHour); // -1 (unlimited) * ``` */ export function loadRateLimitConfig( provider: string, overrides?: Partial<RateLimitSettings> ): RateLimitSettings { const baseConfig = DEFAULT_RATE_LIMITS[provider.toLowerCase()] || DEFAULT_RATE_LIMITS.copilot; return { ...baseConfig, ...overrides, }; } /** * Update rate limit for a provider at runtime * * Dynamically adjusts the rate limit for a specific provider without * restarting the application. Useful for responding to API quota changes * or adjusting limits based on usage patterns. * * Note: This modifies the global DEFAULT_RATE_LIMITS object, so changes * affect all future rate limiter instances for this provider. * * @param provider - LLM provider name (case-insensitive) * @param newLimit - New requests per hour limit (-1 for unlimited) * * @example * ```ts * // Increase OpenAI limit during off-peak hours * updateRateLimit('openai', 5000); * * // Temporarily disable rate limiting for testing * updateRateLimit('copilot', -1); * * // Restore default limit * updateRateLimit('copilot', 2500); * ``` */ export function updateRateLimit(provider: string, newLimit: number): void { if (DEFAULT_RATE_LIMITS[provider.toLowerCase()]) { DEFAULT_RATE_LIMITS[provider.toLowerCase()].requestsPerHour = newLimit; } }