Imported from jeremylongshore/tons-of-skills-marketplace (skills/.curated/salesforce-rate-limits/SKILL.md). Install upstream with npx skills add jeremylongshore/tons-of-skills-marketplace --skill salesforce-rate-limits. Copyright stays with the author (MIT).

Salesforce Rate Limits

Overview

Handle Salesforce API limits gracefully. Salesforce uses a 24-hour rolling limit (not per-minute), plus concurrent request limits and Bulk API quotas.

Prerequisites

• jsforce connection configured
• Understanding of your org's edition and license count
• Access to Setup > Company Information

Instructions

Step 1: Understand Salesforce API Limits

Key difference from most SaaS APIs: Salesforce limits are per-org, not per-user or per-key. All integrations sharing the same org share the same pool.

Step 2: Monitor Remaining Quota

import { getConnection } from './salesforce/connection';

async function checkApiLimits(): Promise<{
  used: number;
  remaining: number;
  max: number;
  percentUsed: number;
}> {
  const conn = await getConnection();
  const limits = await conn.request('/services/data/v59.0/limits/');

  const daily = limits.DailyApiRequests;
  const used = daily.Max - daily.Remaining;
  const percentUsed = (used / daily.Max) * 100;

  return {
    used,
    remaining: daily.Remaining,
    max: daily.Max,
    percentUsed: Math.round(percentUsed * 10) / 10,
  };
}

// Also available in every REST API response header:
// Sforce-Limit-Info: api-usage=135/150000

Step 3: Implement Backoff for REQUEST_LIMIT_EXCEEDED

async function withSalesforceRetry<T>(
  operation: () => Promise<T>,
  config = { maxRetries: 5, baseDelayMs: 2000, maxDelayMs: 60000 }
): Promise<T> {
  for (let attempt = 0; attempt <= config.maxRetries; attempt++) {
    try {
      return await operation();
    } catch (error: any) {
      const errorCode = error.errorCode || error.name;

      // Only retry on transient/limit errors
      const retryable = [
        'REQUEST_LIMIT_EXCEEDED',
        'SERVER_UNAVAILABLE',
        'UNABLE_TO_LOCK_ROW',
      ];

      if (attempt === config.maxRetries || !retryable.includes(errorCode)) {
        throw error;
      }

      // Exponential backoff with jitter
      const exponentialDelay = config.baseDelayMs * Math.pow(2, attempt);
      const jitter = Math.random() * 1000;
      const delay = Math.min(exponentialDelay + jitter, config.maxDelayMs);

      console.warn(`${errorCode}: retry ${attempt + 1}/${config.maxRetries} in ${Math.round(delay)}ms`);
      await new Promise(r => setTimeout(r, delay));
    }
  }
  throw new Error('Unreachable');
}

Step 4: Pre-Flight Limit Check

class SalesforceQuotaGuard {
  private warningThreshold = 0.8; // Warn at 80%
  private blockThreshold = 0.95;  // Block at 95%

  async canMakeRequest(estimatedCalls: number = 1): Promise<{
    allowed: boolean;
    remaining: number;
    reason?: string;
  }> {
    const { remaining, max, percentUsed } = await checkApiLimits();

    if (remaining < estimatedCalls) {
      return {
        allowed: false,
        remaining,
        reason: `Only ${remaining} API calls remain (need ${estimatedCalls})`,
      };
    }

    if (percentUsed / 100 >= this.blockThreshold) {
      return {
        allowed: false,
        remaining,
        reason: `API usage at ${percentUsed}% — blocking to preserve quota`,
      };
    }

    if (percentUsed / 100 >= this.warningThreshold) {
      console.warn(`API usage at ${percentUsed}% (${remaining} remaining)`);
    }

    return { allowed: true, remaining };
  }
}

Step 5: Reduce API Call Count

// STRATEGY 1: Use sObject Collections (1 call = 200 records)
// Instead of 200 individual creates...
const records = contacts.map(c => ({ FirstName: c.first, LastName: c.last, Email: c.email }));
await conn.sobject('Contact').create(records); // 1 API call, not 200

// STRATEGY 2: Use Composite API (1 call = 25 operations)
// See salesforce-core-workflow-b

// STRATEGY 3: Use Bulk API for 10K+ records (1 job = unlimited records)
// Bulk API has its own separate limit pool

// STRATEGY 4: Cache describe calls — metadata rarely changes
const describeCache = new Map<string, any>();
async function cachedDescribe(sObjectType: string) {
  if (!describeCache.has(sObjectType)) {
    describeCache.set(sObjectType, await conn.sobject(sObjectType).describe());
  }
  return describeCache.get(sObjectType);
}

// STRATEGY 5: Use queryMore for pagination (doesn't count as extra API call)
let result = await conn.query('SELECT Id, Name FROM Contact');
while (!result.done) {
  result = await conn.queryMore(result.nextRecordsUrl!);
  // Process result.records
}

Output

• API limit monitoring with threshold alerts
• Automatic retry with exponential backoff for limit errors
• Pre-flight quota checks before batch operations
• Strategies to reduce API call consumption

Error Handling

Examples

Throttle a bulk integration before quota exhaustion

Read the Limits REST resource at the start of each batch, reserve a safety margin agreed with other integrations, and reduce concurrency or defer non-urgent work when the margin is approached. Use exponential backoff with jitter only for transient failures, never for a hard daily quota exhaustion. Test the policy against sandbox limits with synthetic work, record the deferred batch identifiers, and resume only after a fresh quota check confirms capacity.

Resources

• API Request Limits
• Limits REST Resource
• Apex Governor Limits

Next Steps

For security configuration, see salesforce-security-basics.