Skip to main content

Examples

Worked end-to-end examples — copy, adapt, run.

A one-shot run​

./run.sh run "read README.md and summarize it in 5 bullets"

Vanta starts the kernel if needed, reads the file (a kernel-allow read), and prints the summary. Nothing is written, so nothing prompts.

An interactive session​

$ vanta
❯ set a goal: ship the docs site
  ◎ goal set: ship the docs site

❯ what's left to do?
⏺ Read(ROADMAP.md)
  ⎿  …
  Here are the three remaining slices …

❯ commit the changes with a clear message
⏺ Git(commit)
  Do you want to proceed?  ❯ 1 Yes   2 Yes, don't ask again   3 No   4 Never

A read runs immediately; the commit is irreversible, so the kernel escalates it to ask and you get the approval prompt.

Set a goal, get the next step​

❯ /goal improve test coverage on the kernel client
❯ /next
  → "Add a test for the 401-retry path in safety-client.ts …"

A vague goal auto-surfaces one concrete micro-step; the working goal shows in the footer ◇ line.

Schedule a recurring task​

vanta schedule "summarize my unread email and post a brief" --cron "0 8 * * *"
vanta schedule list

Runs every day at 08:00 when the scheduler fires (vanta cron, OS-invoked, or the always-on vanta gateway).

Write a skill​

Skills live at ~/.vanta/skills/<slug>/SKILL.md:

---
name: weekly-review
description: Run my Friday weekly review across goals, money, and radar.
---

1. Read active goals and the Money OS weekly snapshot.
2. List shipped vs slipped slices from the roadmap.
3. Draft three priorities for next week and ask me to confirm.

vanta skill weekly-review runs it; the skill index is injected into the prompt and the body loads on demand.

Mount an MCP server​

.mcp.json in your project root:

{
  "mcpServers": {
    "sqlite": { "command": "npx", "args": ["-y", "@modelcontextprotocol/server-sqlite", "./data.db"] }
  }
}

Its tools mount as kernel-gated Vanta tools on next launch. See MCP integration.

Add a custom tool​

A minimal tool — see Extending Vanta for the full walkthrough:

// tools/word-count.ts
export const wordCount: Tool = {
  schema: { name: "word_count", description: "Count words in a file.",
    parameters: { type: "object", properties: { path: { type: "string" } }, required: ["path"] } },
  describeForSafety: (a) => `read file ${a.path}`,
  async execute(args, ctx) {
    const text = await ctx.readFile(args.path);
    return { ok: true, output: `${text.trim().split(/\s+/).length} words` };
  },
};

Register it in tools/all-tools.ts — it's then kernel-gated like every built-in.

Full reference​