@pi-lab/grok-build

Grok Build CLI tool extension for pi coding agent

Packages

Package details

extension

Install @pi-lab/grok-build from npm and Pi will load the resources declared by the package manifest.

$ pi install npm:@pi-lab/grok-build
Package
@pi-lab/grok-build
Version
1.0.1
Published
Jun 2, 2026
Downloads
365/mo · 27/wk
Author
cheny341
License
MIT
Types
extension
Size
10 KB
Dependencies
0 dependencies · 3 peers
Pi manifest JSON
{
  "extensions": [
    "./dist/index.mjs"
  ]
}

Security note

Pi packages can execute code and influence agent behavior. Review the source before installing third-party packages.

README

@pi-lab/grok-build

A pi extension that adds a single grok_build tool backed by the local Grok Build CLI.

The tool intentionally stays prompt-only: Grok Build is itself an agent, so pi gives it a natural-language task and lets Grok choose its own image, video, web, and X/Twitter capabilities.

Requirements

Install and authenticate Grok Build first.

Capability guidance

Capability Guidance
Image generation ✅ Recommended
Image edit ✅ Recommended
Video generation ✅ Recommended
X/Twitter search ✅ Recommended when Grok-native user, semantic, or advanced keyword search is useful
X/Twitter post/thread read ✅ Use Grok Build for detailed context; use webfetch when faster, simpler output is enough
General web search ⚠️ Available but not recommended; prefer other websearch tool
General web fetch ⚠️ Available but not recommended; prefer other webfetch tool
Coding tasks ❌ Not recommended

Install

pi install npm:@pi-lab/grok-build

Usage

Ask pi for image or video generation, for example:

Use Grok Build to generate an image of a cyberpunk red panda mascot.

The tool accepts only one argument: prompt. Grok decides how to satisfy the request.

X/Twitter usage

grok_build can handle common X/Twitter requests in natural language:

Search X for recent reactions to the latest Grok Build release.
Read this X thread and summarize the main points: https://x.com/...

For X/Twitter links, Grok Build usually returns richer context such as replies, quoted posts, media details, views, and bookmarks. If you only need the text of a single post, webfetch is usually faster and simpler.