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.

Instructions are the agent’s always-on system prompt — its permanent identity rather than a procedure it pulls in when the moment calls for it. eve prepends the instructions to every model call in the session. Use them for anything that should hold on every turn: a persona, a standing rule, a constraint, or a disclosure. They are separate from skills, which are on-demand procedures loaded only when a request calls for them.

​
Authoring instructions

At minimum, instructions are a markdown file at the agent root. Whatever you write becomes the prompt:
agent/instructions.md
You are a concise assistant. Use tools when they are available.
Keep this file to stable, identity-defining content: tone, persona, and standing rules. Long or situational procedures belong in skills, where they only enter context when relevant.

​
Markdown vs TypeScript

A static prompt belongs in a plain markdown file. This is the preferred starting point:
agent/instructions.md
You are a helpful assistant. Always respond in the user's language.
Prefer concise answers and use bullet points for lists.
You cannot author both instructions.md and instructions.ts at the agent root at the same time — that pairing is a build error.

​
Split instructions across a directory

For larger prompts or when different parts of the instructions have different update cadences, add an agent/instructions/ directory. eve reads its entries non-recursively and accepts both .md files and .ts modules. Entries combine in alphabetical order by filename (localeCompare). A flat agent/instructions.md (or .ts) at the agent root and the directory can coexist. The root file’s content comes first, then the sorted directory entries: 
agent/
├── instructions.md          ← comes first
└── instructions/
    ├── 01_persona.md        ← then sorted directory entries
    ├── 02_rules.md
    └── 03_disclaimer.ts

​
Dynamic instructions

To resolve the prompt at runtime from session context — auth, tenant, or channel — wrap defineInstructions in a defineDynamic resolver. This lets you serve a different prompt per principal, organization, or channel without rebuilding the agent:
agent/instructions.ts
import { defineDynamic } from "eve/instructions";
import { defineInstructions } from "eve/instructions";

export default defineDynamic(async (ctx) => {
  const tenant = ctx.session.auth.current?.principalId;
  return defineInstructions({
    markdown: `You are an assistant for ${tenant ?? "a user"}. Follow their team's style guide.`,
  });
});

​
Instructions vs skills

Both instructions and skills feed text into the model’s context. The difference is timing:
LoadedUse for
instructions.md / .tsAlways on, every turnPermanent identity and standing rules
agent/skills/*On demand, when the model calls load_skillOptional procedures that should not bloat every turn
Instructions never run code. When you need typed executable behavior, reach for a tool instead.

​
Subagents

Instructions are required on the root agent. On a declared subagent they are optional — the subagent falls back to the framework default when no instructions are authored: 
agent/subagents/researcher/
├── agent.ts          ← required
└── instructions.md   ← optional on subagents

​
Responsible use

As the deployer, it is your responsibility to ensure your agent complies with applicable laws. Where an eve agent communicates with people, you may be required to disclose that they are interacting with an automated AI system. eve does not add this disclosure automatically — configure it in your instructions and/or channel responses.

Build docs developers (and LLMs) love

Get started for freeTalk to us