tutorial-guide-agent-runtime

Teaches LingTai agent runtime internals including init.json, heartbeat signals, and system prompts.

692|60|Updated Mar 15, 2026
One-click install
npx skills add https://github.com/Lingtai-AI/lingtai --skill tutorial-guide-agent-runtime
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: tutorial-guide-agent-runtime
Source: https://github.com/Lingtai-AI/lingtai/tree/main/tui/internal/preset/skills/lingtai-tutorial-guide/reference/agent-runtime
Command: npx skills add https://github.com/Lingtai-AI/lingtai --skill tutorial-guide-agent-runtime

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Understanding how a LingTai agent actually boots, runs, and forms its identity requires reading scattered files and source code; this Skill guides live, evidence-based teaching of agent runtime internals for tutorial lessons 4–6.

Core Features & Use Cases

  • init.json Walkthrough: Reads the real init.json and explains fields like manifest, addons, env_file, and the _file convention instead of reciting documentation.
  • Runtime & Signal Inspection: Demonstrates the lingtai-agent run boot sequence, heartbeat files, and signal files (.interrupt, .suspend, .sleep, .prompt) using live examples.
  • System Prompt Discovery: Runs a Python snippet against SystemPromptManager to reveal the actual section render order and explains protected versus editable identity sections.
  • Use Case: A new LingTai user wants to understand why their agent persists across sessions and how its personality is stored; the guide reads the user's own system/lingtai.md and heartbeat files to explain concretely.

Quick Start

Ask the agent to walk you through how it was born by reading its own init.json and explaining each field it finds.

Frequently Asked Questions about tutorial-guide-agent-runtime

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

FAQPage Schema
How do I run a LingTai agent without the TUI?

A valid init.json plus the lingtai-agent run command is sufficient to start an agent. The TUI is only a Go frontend that writes init.json, launches the Python process, and monitors heartbeat files.

How does the LingTai agent boot sequence work?

The agent reads init.json, loads the env file, resolves the venv, builds the Agent object, cleans stale signals, installs signal handlers, starts in ASLEEP state, and blocks on shutdown. It is a long-running Python process that does not exit after one task.

What are LingTai signal files like .interrupt and .suspend?

Signal files (.interrupt, .suspend, .sleep, .prompt) are filesystem-based controls placed in the agent's home directory. External tools like the TUI write them to control the running agent, while .agent.heartbeat tracks liveness with updating timestamps.

Can an agent edit its own system prompt sections?

Only editable sections like identity/lingtai and pad can be changed by the agent; protected sections such as principle, covenant, rules, and procedures are host-written. The brief is externally maintained by the secretary.

Where is a LingTai agent's personality stored?

Character is durable state the agent authors for itself in system/lingtai.md via the psyche, not a seed field in init.json. This identity section is how agents develop unique personalities through experience.