headless

Control a headless ikigai instance by sending keystrokes and reading the framebuffer.

Updated Mar 4, 2026

Search Tags:#automation #cli #llm #headless #agent-session #ikigai #framebuffer

Ready to run

npx skills add https://github.com/multicam/ikigai-rev --skill headless-multicam

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: headless
Source: https://github.com/multicam/ikigai-rev/tree/main/.claude-i/library/headless
Command: npx skills add https://github.com/multicam/ikigai-rev --skill headless-multicam

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Enables controlling a running ikigai instance in headless mode from an agent session, allowing automated interaction without a visible terminal.

Core Features & Use Cases

Headless launch and background operation with ikigai --headless
Interact-Observe Loop: send keystrokes, wait for UI updates, read the framebuffer
Read framebuffer to interpret screen output (lines and spans)
Timing and synchronization to coordinate prompts and responses

Quick Start

Start ikigai in headless mode and drive interaction with the LLM using ikigai-ctl commands.

Frequently Asked Questions about headless

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

FAQPage Schema

How do I control an LLM CLI in headless mode from an automated agent session?▼

You can control an LLM CLI in headless mode by launching the instance with a headless flag and using a control utility to simulate keystrokes, read the framebuffer, and coordinate response timing within an agent session.

What is the interact-observe loop for headless CLI automation?▼

The interact-observe loop is an automation pattern where you send simulated input keystrokes to a background process, wait for UI updates, and read the framebuffer output to interpret the resulting screen state.

How do I read framebuffer output to interpret screen lines and spans during automation?▼

Reading framebuffer output involves accessing the background process screen buffer to extract text lines and spans, allowing an agent session to interpret visual UI updates and coordinate response timing without a visible terminal.

Can I run background LLM workflows without a visible terminal for automated testing?▼

Yes, you can run LLM workflows without a visible terminal by launching the instance in headless mode, enabling automated testing workflows that simulate input and observe framebuffer output in the background.

What's the best way to synchronize prompt and response timing in a headless LLM deployment?▼

Synchronizing prompt and response timing in a headless LLM deployment requires using timing controls within the interact-observe loop to coordinate when to send simulated input and when to read the framebuffer output.

Why does my headless CLI automation fail to capture screen output correctly?▼

Headless CLI automation may fail to capture screen output if the interact-observe loop timing is misconfigured, causing the framebuffer to be read before the background process finishes updating its UI state.