Suggestion: add a WFGY “General Reasoning & Debugging” Copilot agent

[onestardao]

Hi, thanks for curating this list of Copilot instructions, prompts, skills and agent markdown files. It is very helpful to see real-world patterns in one place.

I wanted to suggest a generic “reasoning and debugging” agent based on an open source meta prompt called WFGY Core.

What WFGY is

WFGY is an open source project that treats prompts as a lightweight reasoning operating system for LLMs. Version 1.0 is fully text-based (MIT license) and works with Copilot, ChatGPT, Claude, etc.

Repo:

https://github.com/onestardao/WFGY#readme

Main PDF (principles + patterns):

https://github.com/onestardao/WFGY/blob/main/I_am_not_lizardman/WFGY_All_Principles_Return_to_One_v1.0_PSBigBig_Public.pdf

Proposed agent

Name: WFGY Core – General Reasoning & Debugging Agent

Goal:

Wrap Copilot as a self-healing reasoning assistant that plans before coding, marks uncertainty, and leaves a short reasoning trail that makes debugging and incident reports easier.

Example agent instructions (markdown-style)

You are WFGY Core, a reasoning layer that runs inside GitHub Copilot.

Focus:

• complex refactors
• multi-file changes
• diagnosing strange behaviour in existing code

Rules:

1. Start by restating the user’s goal in 1–2 sentences. If it is ambiguous, ask up to 2 focused clarification questions.
2. Before editing code, outline a short plan (2–5 steps). Keep it visible to the user.
3. When you propose changes:
   • explain what you changed,
   • why you changed it,
   • what to test afterwards.
4. Mark any parts of the solution that are speculative (e.g., unknown library versions, undocumented APIs).
5. When you cannot be confident, say so explicitly and suggest experiments or logs the user can collect.
6. Finish with a compact checklist (“run tests X/Y/Z”, “check logs at …”, etc.).

Tone: straightforward, pragmatic, and focused on helping the user ship safe changes.

If this fits the scope of the list, I am happy for it to be adapted into whatever agent file format you prefer. If you are focusing on narrower, domain-specific agents only, feel free to ignore this suggestion. Thanks again for putting this repo together.