Protocol, Skill, and Web Audit Templates: The Three Things

Table of Contents

• What This Page Solves
• Shortest Definition First
• Asset Overview
• Part 1: What Is the Protocol
• Part 2: What Is Skill
• Part 3: What Are Web Audit Templates
• Why Web Audit Templates Must Stay Separate from Skill
• Three Common Misreadings
• Recommended Next Steps

What This Page Solves

This page belongs to Layer 3: Relation and Adoption.

It does not answer "how do I run the loop right now?" It answers a different question that often gets mixed up before any comparison can even begin:

• what this repo actually ships
• what Protocol / Skill / Web audit assets each are
• which of these are methodologies, and which are only carriers or stabilizers when you compare them with mainstream approaches

When many people first see this project, they collapse three different things into one:

• They treat the protocol as a package of installable prompts
• They treat Skill as the protocol itself
• They treat Web audit templates as local capabilities that also need installation

All three misunderstandings send the entry logic in the wrong direction.

Shortest Definition First

Cyber-Ming-Protocol is a protocol-first project.

• Protocol defines the governance skeleton
• Skill stabilizes high-frequency actions on the IDE side
• Web audit templates stabilize audit questions on the Web side

All three serve the same protocol, but they are not the same kind of thing.

The safest public description is: Protocol / Skill / Docs are the project's three main delivery forms, while web-audit-templates/ is a separate collaboration asset for the Web side.

This page focuses on protocol, Skill, and templates because many comparisons start by comparing the wrong layer: they mix delivery forms with methodologies, and everything after that becomes muddy.

Asset Overview

Object	What It Is	Where It Lives	Main Function	Required	Often Mistaken For
Protocol	A set of human-AI governance rules	README.md, wiki/	Defines what must happen first, what counts as completion, and who is responsible for what	Yes, you must understand it	A prompt pack or a stylistic narrative
Skill	A stable trigger skeleton on the IDE side	skill/	Helps you trigger planning, execution, probing, and renewal more reliably	No, you can install it later	The protocol itself or an automatic judge
Docs / Teaching	The explanatory and teaching layer of the project	README.md, wiki/	Helps you understand the protocol, cases, boundaries, and how to get started	Yes, you need enough of it to understand the system	Disposable supporting pages
Web Audit Templates	Audit collaboration templates for the Web side	web-audit-templates/	Helps you do plan audit, completion audit, and renewal judgment more reliably	No, optional	A local Skill or an automatic executor

Part 1: What Is the Protocol

The protocol is the truly irreplaceable part of this project.

It defines a governance structure such as:

• Review first, then execute
• Submit the Atomic Execution Contract first, then allow work to begin
• The executor does not self-certify completion
• Completion must be established by logs, artifacts, run results, and commits as evidence
• When a window decays, you use renewal instead of dragging the old context forward

All of that still holds even if you never install Skill.

Part 2: What Is Skill

Skill does not invent the protocol for you. Its role is to help you hold the protocol more reliably on the IDE side.

For example, the skills in this repo solidify several high-frequency actions:

• approval-first-planner: Produce the Atomic Execution Contract and boundaries for approval first
• approved-checklist-executor: Execute, verify, and archive strictly by approved slices
• probe-first-scout: Do not pretend to understand the whole system; run the smallest probe first
• legacy-project-handover: Provide a read-only snapshot for takeover or renewal
• auditor-succession-prompt: Package a clean renewal packet for a fresh Web auditor seat
• free-development-mode: Continue long-running development under explicit red-line control after the user deliberately activates it

But they do not change one thing:

Skill cannot define truth for you.

But it can now do one more important thing:

it can bootstrap and maintain the place where current truth should live.

When serious work begins, the corresponding skill layer can bootstrap dev_repo/ so that state.json, journal.jsonl, evidence_index.json, and tree.md hold current runtime truth in repository artifacts rather than only inside chat residue.

That does not make Skill the judge of truth. It means Skill can create and maintain the runtime container where auditable truth is recorded.

Part 3: What Are Web Audit Templates

Web audit templates are not local Skill, and they are not IDE-side triggers.

They simply help you play the auditor role more reliably on the Web side:

• Audit whether the plan is atomic enough
• Audit whether completion is real
• Audit whether the current window has decayed and needs renewal

They provide the scaffold for how to ask, how to inspect, and how to judge, not for how to execute.

Why Web Audit Templates Must Stay Separate from Skill

Because the two sides hold different roles:

• IDE-side Skill maps to execution, planning, probing, and takeover
• Web-side templates map to plan audit, completion audit, and renewal judgment

If you put everything into skill/, readers will easily misunderstand three things:

• That Web audit is also a local installable
• That executor and auditor are just the same kind of tool with different prompts
• That audit can be completed conveniently inside the same window

That would weaken the most important layer of the protocol: role separation and sovereignty in human hands.

Three Common Misreadings

Misreading 1: Installing Skill Makes the Protocol Work Automatically

Wrong. Installing Skill only makes the right actions easier to trigger. It does not mean you already satisfy the protocol's evidence standard.

Misreading 2: Without Skill, You Cannot Use Cyber-Ming

Wrong. This protocol is first of all a governance method that can be executed by hand. You can absolutely run the loop manually first and decide about Skill later.

Misreading 3: Web Audit Templates Are Also Skills You Install Locally

Wrong. Web templates help the auditor question, verify, and judge. They are not IDE-side local skills.

Recommended Next Steps

• If you want to stay inside Layer 3, continue with Comparison and map this against mainstream approaches
• If you have not yet completed Layer 1, go back to Minimal Loop Guide
• If you have already completed Layer 1 and now want Layer 2, go back to Minimal Stable Loop Guide