AgentSkillsCN

lindy-sdk-patterns

Lindy AI SDK最佳实践和常见模式。 在学习SDK模式、优化API使用或实现高级代理功能时使用。 可通过“lindy SDK patterns”、“lindy best practices”、“lindy API patterns”、“lindy code patterns”等短语触发。

SKILL.md
--- frontmatter
name: lindy-sdk-patterns
description: |
  Lindy AI SDK best practices and common patterns.
  Use when learning SDK patterns, optimizing API usage,
  or implementing advanced agent features.
  Trigger with phrases like "lindy SDK patterns", "lindy best practices",
  "lindy API patterns", "lindy code patterns".
allowed-tools: Read, Write, Edit
version: 1.0.0
license: MIT
author: Jeremy Longshore <jeremy@intentsolutions.io>

Lindy SDK Patterns

Overview

Essential SDK patterns and best practices for Lindy AI agent development.

Prerequisites

  • Completed lindy-install-auth setup
  • Basic understanding of async/await
  • Familiarity with TypeScript

Instructions

Pattern 1: Client Singleton

typescript
// lib/lindy.ts
import { Lindy } from '@lindy-ai/sdk';

let client: Lindy | null = null;

export function getLindyClient(): Lindy {
  if (!client) {
    client = new Lindy({
      apiKey: process.env.LINDY_API_KEY!,
      timeout: 30000,
    });
  }
  return client;
}

Pattern 2: Agent Factory

typescript
// agents/factory.ts
import { getLindyClient } from '../lib/lindy';

interface AgentConfig {
  name: string;
  instructions: string;
  tools?: string[];
}

export async function createAgent(config: AgentConfig) {
  const lindy = getLindyClient();

  const agent = await lindy.agents.create({
    name: config.name,
    instructions: config.instructions,
    tools: config.tools || [],
  });

  return agent;
}

Pattern 3: Retry with Backoff

typescript
async function runWithRetry<T>(
  fn: () => Promise<T>,
  maxRetries = 3
): Promise<T> {
  for (let i = 0; i < maxRetries; i++) {
    try {
      return await fn();
    } catch (error: any) {
      if (error.status === 429 && i < maxRetries - 1) {
        await new Promise(r => setTimeout(r, Math.pow(2, i) * 1000));
        continue;
      }
      throw error;
    }
  }
  throw new Error('Max retries exceeded');
}

Pattern 4: Streaming Responses

typescript
async function streamAgentResponse(agentId: string, input: string) {
  const lindy = getLindyClient();

  const stream = await lindy.agents.runStream(agentId, { input });

  for await (const chunk of stream) {
    process.stdout.write(chunk.delta);
  }
  console.log(); // newline
}

Output

  • Reusable client singleton pattern
  • Agent factory for consistent creation
  • Robust error handling with retries
  • Streaming support for real-time output

Error Handling

PatternUse CaseBenefit
SingletonConnection reuseReduced overhead
FactoryAgent creationConsistency
RetryRate limitsReliability
StreamingLong responsesBetter UX

Examples

Complete Agent Service

typescript
// services/agent-service.ts
import { getLindyClient } from '../lib/lindy';

export class AgentService {
  private lindy = getLindyClient();

  async createAndRun(name: string, instructions: string, input: string) {
    const agent = await this.lindy.agents.create({ name, instructions });
    const result = await this.lindy.agents.run(agent.id, { input });
    return { agent, result };
  }

  async listAgents() {
    return this.lindy.agents.list();
  }

  async deleteAgent(id: string) {
    return this.lindy.agents.delete(id);
  }
}

Resources

Next Steps

Proceed to lindy-core-workflow-a for agent creation workflows.