Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/vercel/eve/llms.txt

Use this file to discover all available pages before exploring further.

eve is a framework for building durable AI agents as ordinary files in a TypeScript project. Instead of one large configuration object, each part of your agent gets a clear, predictable home: instructions go in one file, tools in one folder, channels in another. eve discovers that structure, turns it into a running agent, and keeps every session alive across restarts, redeploys, and multi-turn conversations — without you writing any of that machinery yourself.

​
An eve project at a glance

A small eve app looks like this: 
my-agent/
├── package.json
└── agent/
    ├── agent.ts
    ├── instructions.md
    ├── tools/
    │   └── get_weather.ts
    ├── skills/
    │   └── plan_a_trip.md
    └── channels/
        └── slack.ts
You can understand most eve projects by reading that tree:
  • instructions.md — the always-on system prompt that tells the agent who it is and how to behave.
  • agent.ts — chooses the model and configures runtime options.
  • tools/ — typed functions the model can call.
  • skills/ — longer procedures the model loads only when they are useful.
  • channels/ — connects the agent to HTTP clients, Slack, Discord, and other messaging surfaces.
Start with only instructions.md and agent.ts. Add the other folders as the agent needs them.

​
The filesystem is the interface

eve is filesystem-first. A file’s location says what it does, and its path usually gives it a name. For example, this file: 
agent/tools/get_weather.ts
defines a tool named get_weather:
agent/tools/get_weather.ts
import { defineTool } from "eve/tools";
import { z } from "zod";

export default defineTool({
  description: "Get the weather for a city.",
  inputSchema: z.object({ city: z.string() }),
  async execute({ city }) {
    return { city, condition: "Sunny" };
  },
});
There is no separate registry to keep in sync. Add the file and eve discovers it automatically; move or rename it and its identity moves with it.

​
What happens when a message arrives

The same flow runs whether a message comes from a web app, the terminal, or Slack:
  1. eve turns the platform input into a message.
  2. The model receives its instructions, skills, tools, and conversation history.
  3. eve runs the work — calling tools and subagents as needed.
  4. Progress is saved and events are streamed in real time.
  5. The result is delivered back in the form the platform expects.
That keeps agent behaviour portable. Your weather tool does not need to know whether the question came from a browser or from Slack.

​
Durable by default

An eve session is more than one request and one response. A session can:
  • Stream progress while work is happening.
  • Call tools and subagents.
  • Pause for approval or a human answer.
  • Resume after that answer arrives — even much later.
  • Keep durable state across many turns.
Under the hood, eve uses the open-source Workflow SDK to make sessions durable, resumable, and crash-safe. eve handles that machinery so your tools can focus on the work itself.
Completed steps within a turn never re-run; eve replays the recorded result. A step interrupted mid-execution re-runs, so make non-idempotent side effects (charges, emails) idempotent, or gate them with an approval step.

​
Grow the project by adding capabilities

As the agent grows, each new concern still has a predictable home:
PathAdd it when you need…
connections/Tools from external MCP servers
hooks/Code that reacts to lifecycle and stream events
sandbox/A controlled workspace for files and commands
subagents/Specialist agents the root agent can delegate to
schedules/Recurring or scheduled work
lib/Shared code imported by the other agent files
The result stays readable before it runs. The directory tells you what the agent can do.

Quickstart

Scaffold and run your first working agent in minutes.

Tools

Define typed actions the model can call, with Zod schemas and full TypeScript inference.

Instructions

Shape agent behaviour with the always-on system prompt.

Channels

Reach the agent from Slack, Discord, or a custom web UI.

Build docs developers (and LLMs) love

Get started for freeTalk to us