AI Protocol Kit is a collection of behavioral contracts you give to an AI before it works. Each protocol tells the AI how to approach a task, what to check, what not to assume, when to stop, and what kind of output is acceptable. These are not magic prompts — they are working constraints shaped from real, messy workflows.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.
Introduction
Understand what protocols are, why they exist, and how to choose the right one for your situation.
How to Use
Learn the four steps for loading a protocol and putting it to work in any AI session.
Protocol Map
Browse all 17 protocols by lane — match your failure mode to the right contract.
Minimal Stacks
Pre-built protocol combinations for the most common workflows: GitHub publishing, page prep, and more.
Pick by failure mode, not by name
The fastest way to choose a protocol is to identify what you want to prevent. Each lane below maps a common failure mode to the protocols that address it.Shape the Work
Idea too fuzzy? Task collapsing too quickly? Use these before execution starts.
Brief the Artifact
Need a complete brief before writing, building, or structuring output? Start here.
Review and Findings
Inspecting, reviewing, or capturing bugs? Keep evidence, inference, and conclusions separated.
GitHub and Publication
Publishing a repo, README, Pages site, or public page? Close all risks before you push.
Agentic Sessions
Running complex AI sessions across files, tools, or multiple passes? Give the AI stronger rails.
Default rule: use one protocol
Stack protocols only when one controls the session and another controls the artifact or publication target. The Minimal Stacks section has pre-built combinations for the most common scenarios — use those before composing your own.These protocols do not guarantee correctness. They reduce common failure modes: rushing to the obvious answer, writing for the wrong reader, inventing unsupported claims, publishing before review, or letting AI tools approve their own work.