🪙 Skill: Omni Context Minimizer

⚠️ When to Trigger this Skill?

As a Subagent, you are PROHIBITED from reading massive code files (.js, .py, .go, .ts) with a full-file Read without first going through this minimizer, unless you already know exactly the line number or range you are going to inspect.

If you are tasked to "Analyze how the API is structured" or "Review the large views.py file":

1. YOU CANNOT LOAD THE ENTIRE FILE INTO YOUR MEMORY.
2. EXECUTE THIS SKILL IMMEDIATELY.

🛠️ How it Works (Instructions)

This repository contains a Python script called omni_minimizer.py in the scripts/ folder. This script scrutinizes the lexical DOM of the target file and returns only the lines that declare Functions, Classes, or Imports via the terminal.

Step 1: Locate the absolute path of the heavy file you need to analyze (e.g., /path/to/my_project/src/app.js).

Step 2: Run the minimizer passing that path as an argument (the Python script must be executed from the Skill's path in the submodule):

python .agents/skills/omni-context-minimizer/scripts/omni_minimizer.py /path/to/heavy/file

Step 3: The console will return the skeleton. For example:

Line 1: import express from 'express';
Line 4: const app = express();
Line 10: export const loginController = async (req, res) => {
Line 45: class DatabaseService {
...
--- [OPTIMIZATION]: The file was reduced to 8 lines of pure structure. ---

Step 4: Now that you have the "map", if you decide you need to see how the logic of loginController at line 10 works, you can use a targeted Read (offset/limit) or ripgrep focusing your shot STRICTLY on line 10, saving all the remaining context.

🔴 Token-Saver Authorization

If you are sending an implementation_plan.md to the Orchestrator and it includes the heuristic scanning of an entire /src folder, you MUST specify in your MD: "The omni-context-minimizer will be used to map the 5 base files, preventing context drowning". Otherwise, the Token-Saver Auditor will reject the plan.