Why does this skill refer to the user as 'Q'?

The protocol uses 'Q' as a standardized shorthand for the human collaborator to maintain a clear distinction between the agent's internal reasoning and its communication with the user.

How does the Explicit Reasoning Protocol work?

Before every action, Claude writes a specific predicted outcome (EXPECT). After the action, it compares the result to the prediction (MATCHES) and concludes the next step based on evidence.

Will this skill make Claude work slower?

Yes, intentionally. It follows the 'slow is smooth, smooth is fast' philosophy, prioritizing accuracy and error prevention over speed to avoid the high cost of compounding mistakes.

What is the 'One Rule' of this protocol?

The core rule is that reality takes precedence over the agent's model. When reality contradicts the AI's internal map, the agent must stop and fix its model before proceeding.

When should I use the General Agentic Protocol?

It is best used for complex programming tasks, system-level changes, and debugging sessions where the cost of being wrong exceeds the cost of being slow.

General Agentic Protocol

Name: General Agentic Protocol
Author: CodingKaiser

byCodingKaiser

0•

生産性とワークフロー

Implements a defensive epistemology framework for AI coding agents to minimize errors through explicit prediction and reality verification.

The General Agentic Protocol is a specialized skill that applies rationalist principles to the coding process, transforming how Claude interacts with software environments. Instead of proceeding on assumptions, this protocol mandates a scientific 'predict-act-verify' loop that forces the agent to acknowledge confusion and update its mental model whenever reality contradicts its expectations. By establishing rigorous evidence standards, root cause discipline, and clear autonomy boundaries, this skill ensures that Claude operates with high reliability in complex codebases where compounding mistakes can be catastrophic. It is designed for high-stakes development where precision and clear communication with the human collaborator are paramount.

主な機能

010 GitHub stars

02Checkpoint-Based Feedback: Limits action batches to ensure continuous alignment with observable reality.

03Explicit Reasoning Protocol: Prediction, Action, and Verification cycles for every tool call.

04Root Cause Discipline: Mandatory 'Five Whys' analysis to solve systemic issues rather than surface symptoms.

05Handoff Protocol: Maintains clean project states and clear documentation for future sessions.

06Confusion Awareness: Triggers immediate model updates when reality diverges from expectations.

ユースケース

01Managing complex refactors where hidden dependencies could cause cascading failures.

02Debugging non-obvious errors by maintaining and testing multiple competing hypotheses.

03Executing high-risk operations like database schema changes or critical API updates.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add codingkaiser/kaiser-skills general-agentic

For use in Claude.ai and ChatGPT

Download Skill