inngest-agents

Creates durable AI agent workflows with in-app human-in-the-loop approvals using InfluxDB and AgentKit.

1|Updated Jun 23, 2026
One-click install
npx skills add https://github.com/YOYOMAII/foundry --skill inngest-agents-yoyomaii
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: inngest-agents
Source: https://github.com/YOYOMAII/foundry/tree/main/.agents/skills/inngest-agents
Command: npx skills add https://github.com/YOYOMAII/foundry --skill inngest-agents-yoyomaii

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires @inngest/agent-kit, inngest.

What problem does it solve?

This skill solves the fragility of standard AI agent loops by providing a framework for durable, long-running workflows that survive server restarts, API failures, and complex multi-step processes.

Core Features & Use Cases

  • Durable Execution: Ensures agent state, tool calls, and model inferences are persisted, allowing workflows to resume after crashes or timeouts.
  • Human-in-the-Loop: Integrates native support for pausing execution to wait for human approval or external signals.
  • Realtime Progress: Streams agent status and progress updates directly to the UI from within the durable workflow.
  • Use Case: Build a support agent that processes tickets, calls external tools to check database status, waits for a manager's approval via email, and then sends a final response, all without losing state if the server deploys mid-process.

Quick Start

Use the inngest-agents skill to architect a new durable support agent that includes a human-in-the-loop approval step and tool-calling capabilities.

Frequently Asked Questions about inngest-agents

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I build durable AI agent workflows that survive server restarts?

Durable AI agent workflows persist state, tool calls, and model inferences using Inngest step primitives, allowing multi-step processes to resume automatically after crashes or server restarts without losing execution progress.

Can I pause an AI agent workflow to wait for human approval?

You can pause durable AI agent workflows to wait for human-in-the-loop approvals or external signals. The execution state is safely persisted until the approval is received, preventing timeout or state loss.

What's the best way to manage complex backend automation with multi-step tool calls?

Orchestrating complex backend automation with AgentKit manages multi-step tool calls and model inference sequentially. This approach provides fault tolerance and reliable side-effect management for robust execution.

Do I need Inngest and AgentKit to run crash-safe agent loops?

Yes, both Inngest and AgentKit are required dependencies. Integrating Inngest step primitives with AgentKit provides the necessary state persistence and fault tolerance to execute crash-safe agent loops reliably.

How does an AI agent handle state persistence during long-running tool calling?

State persistence during long-running tool calling is handled by saving agent state and model inferences at each step. This ensures workflows survive API failures and can resume exactly where they left off.

Can I stream realtime progress updates from a durable workflow to the UI?

Durable workflows can stream realtime agent status and progress updates directly to the UI. This allows you to monitor multi-step tool calls and human-in-the-loop approvals as they execute.