Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/XxYouDeaDPunKxX/ai-protocol-kit/llms.txt

Use this file to discover all available pages before exploring further.

AI Protocol Kit is a set of practical behavioral contracts you load into an AI context before it starts working. Each protocol tells the AI how to approach a task, what to verify, when to stop, and what kind of output is acceptable — replacing vague prompts with explicit operating rules. These are not magic prompts. They are working contracts for real, messy situations where the AI should not simply improvise.

Introduction

Understand what protocols are, why they exist, and which failure modes they address.

How to Use

How to select, load, and apply a protocol in an AI session — in three steps.

Protocol Map

Choose the right protocol by situation: shaping, briefing, reviewing, publishing, or governing.

Stacking Guide

When and how to combine protocols — and when a single protocol is always the right answer.

Browse by situation

Shape & Reason

Protocols for fuzzy ideas, ambiguous tasks, and pre-execution reasoning. Use these before the work starts.

Brief & Output

Protocols for writing to real readers, briefing HTML tools, and capturing structured findings.

GitHub & Publication

Protocols for repo publication, README authoring, GitHub Pages, badges, and discovery infrastructure.

Agentic & Orchestration

Protocols for governing agentic AI sessions, multi-AI orchestration, and constraint-aware reasoning.

What this kit prevents

Premature convergence

AI rushing to the obvious answer before exploring the problem space.

Invented facts

AI presenting inferred or hallucinated claims as verified findings.

Wrong reader fit

Writing for the wrong audience or tone without closing the reader model first.

Repo publication errors

Committing, pushing, or publishing before workspace and risk are audited.

Chat-summary drift

Treating chat summaries as repository truth rather than reading the actual files and repo state.

Agentic drift

Multi-step AI sessions expanding scope or approving their own work without human checkpoints.

Flat compromises

Constraint conflicts resolved by splitting the difference rather than surfacing the dominant force.
Default rule: use one protocol. Stack only when one protocol governs the session and another governs the artifact or publication target. See the Stacking Guide for justified combinations.

Build docs developers (and LLMs) love