BitFrog Copilot

7+1 AI development agents for GitHub Copilot, powered by Chinese philosophy thinking models.

BitFrog Copilot brings structured, disciplined AI development workflows to GitHub Copilot — not through rigid rules, but through deeply integrated philosophical thinking patterns that guide how each agent reasons, decides, and acts.

Why Philosophy Instead of Rules?

Most AI coding agents use hard rules to enforce discipline: iron laws, hard gates, rationalization blockers. This works — but we found it has limits.

When you force an LLM with "you MUST do X" and "you MUST NOT do Y", the model follows the letter but misses the spirit. It checks boxes without understanding why. The result looks correct but feels mechanical.

Chinese philosophy offers a different approach: instead of constraining behavior from the outside, it shapes how the agent thinks from the inside. An agent that truly understands why tests matter will write them naturally — no iron law needed.

This project is deeply inspired by Superpowers by Jesse Vincent. We are grateful for its foundational concepts and workflow design. BitFrog builds on that foundation by replacing external constraints with philosophical thinking models — not because rules are wrong, but because we believe understanding runs deeper than compliance.

Install

Via VS Code Marketplace: Search "BitFrog Copilot" in Extensions → Install

Via Agent Plugin: Extensions sidebar → Agent Plugins → Search "bitfrog-copilot"

Agents

Agent	Role	Philosophy
@bitfrog	Main router	Diagnose intent before routing
@bitfrog-brainstorm	Explore & design	格物致知 — understand the real problem before proposing solutions
@bitfrog-plan	Map dependencies & decompose tasks	格物 + 辨证 — scout the terrain, then plan
@bitfrog-execute	TDD implementation	知行合一 — if you know you should write tests, write them
@bitfrog-debug	Diagnose & fix	辨证论治 — diagnose the root level, then treat
@bitfrog-review	Two-phase review & completion	三省吾身 — self-reflect, peer-reflect, user-reflect
@bitfrog-mentor	Guided learning	格物致知 as a process — guide, don't tell
@bitfrog-ui-design	UX research	Understand the user's real need before designing the interface

Philosophy

Five principles drive all BitFrog agents (see bitfrog-philosophy.md):

Meta-principle: 中庸之道 — Every action has a right measure. Too much is as bad as too little. This is not "find the middle ground" — it is the judgment to know what is enough in each specific situation.

Core:

格物致知 — Investigate the true nature of things before acting. The user may not know what they really need.
知行合一 — Knowing and doing are inseparable. If you skip what you know you should do, you don't truly know it.
辨证论治 — The same symptom can have different root causes. Diagnose the level of the problem before choosing a fix.

Collaboration: 4. 阴阳互生 — Each agent does its own job while staying aware of the whole system. 5. 三省吾身 — Quality comes from reflecting on your thinking process, not from checking boxes.

Why Not Just Use Rules?

Rules-based (external constraint)	Philosophy-based (internal drive)
"You MUST write tests"	Agent understands why tests matter → writes them naturally
"You MUST NOT skip root cause analysis"	Agent understands that treating symptoms creates more problems
Lists 11 excuses and blocks them	When understanding is deep, excuses don't arise
Checklist: ✅ / ❌	Reflection: "Is my thinking process right?"
Finds the optimal solution	Finds the appropriate solution for this context

Workflow

brainstorm → plan → execute → review
                       ↕
                     debug

Each arrow is a handoff button — agents suggest the next step, you decide when to proceed.

debug and mentor work independently at any time.

Language

All agents support English and 简体中文 automatically.

Acknowledgments

BitFrog Copilot is inspired by Superpowers by Jesse Vincent. The workflow structure, skill-based architecture, and many foundational concepts originate from that project. We are grateful for the open-source foundation it provides.

License

MIT

BitFrog Copilot is part of the BitFrog product family.

BitFrog Copilot

Rain Lei

BitFrog Copilot

Why Philosophy Instead of Rules?

Install

Agents

Philosophy

Why Not Just Use Rules?

Workflow

Language

Acknowledgments

License