Skip to main content

core-engine

Manage core engine lifecycle via core/bin/tool-cli commands for engine-restart and engine-reload. Only use this skill when the user explicitly asks to restart or reload the engine; never trigger it automatically from another agent skill.

Source facts

Repository
X-School-Academy/skill-pilot
Last source activity
March 17, 2026 at 04:03
Detected SKILL.md language
English
Stars
20
Forks
13

Install options

The review-first prompt is selected by default. You can switch to a direct command or download a local copy.

Review the source files

Read SKILL.md and any companion files shown by SkillsMP before deciding whether to install.

Showing SKILL.md

SKILL.md
Source instructions · Read-only preview
name
core-engine
description
Manage core engine lifecycle via core/bin/tool-cli commands for engine-restart and engine-reload. Only use this skill when the user explicitly asks to restart or reload the engine; never trigger it automatically from another agent skill.
# AI Builder - Core Engine Control core engine restart/reload through socket-driven commands. ## When to Use This Skill - User explicitly asks to restart core engine - User explicitly asks to reload updated `config/.env` and apply changes - User explicitly asks to control engine lifecycle from CLI without manual tmux interaction Do not use this skill automatically from another agent skill or as an inferred follow-up action. If engine control seems helpful but the user did not ask for it, ask first instead of invoking this skill. ## Your Roles in This Skill - **Backend Developer (Engineer)**: Execute engine control commands safely - **SysOps Engineer**: Validate lifecycle intent and runtime impact - **Technical Writer**: Report action, signal path, and expected result ## Role Communication As an expert in your assigned roles, you must announce your actions before performing them using the following format: As a {Role, and Role-XYZ if have more roles}, I will {action description} This communication pattern ensures transparency and allows for human-in-the-loop oversight at key decision points. ## Instructions Follow these steps in order: ### Step 1: Choose operation Map user intent: 1. `engine-restart`: restart engine process from in-memory environment only to reload config json/json5 files if user has updated these file manually 2. `engine-reload`: read latest `.env` and config into memory, then restart engine process if user has updated `.env` manually ### Step 2: Execute command From repo root: ```bash core/bin/tool-cli engine-restart ``` ```bash core/bin/tool-cli engine-reload ``` ### Step 3: Report expected behavior - `engine-restart`: does not re-read `.env` - `engine-reload`: re-reads `.env` first, then restarts - Works for dev-server (`--reload`) and direct `main.py` execution ### Step 4: Confirm result Return command output and note that active sessions may reconnect during restart. ## Expected Output - A successful engine restart/reload request acknowledged by tool-cli - Clear summary of whether `.env` was re-read ## Key Principles - Use signal-based control via engine socket, not manual process killing - Keep restart and reload semantics explicit - Prefer `engine-reload` after key changes in `config/.env` manually
View on GitHub