omk-global-rules

Enforce global operating rules for multi-step coding workflows.

130|14|Updated Apr 30, 2026
One-click install
npx skills add https://github.com/dmae97/oh-my-kimi --skill omk-global-rules
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: omk-global-rules
Source: https://github.com/dmae97/oh-my-kimi/tree/main/templates/skills/kimi/omk-global-rules
Command: npx skills add https://github.com/dmae97/oh-my-kimi --skill omk-global-rules

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It solves inconsistent agent behavior during multi-step coding or coordination by enforcing a shared set of operational rules for how to plan, verify, report changes, and use supporting tools or subagents.

Core Features & Use Cases

  • Evidence-gated and checkpointed transitions: Uses checkpoint preferences (e.g., Kimi Okabe + SendDMail) for risky context changes and rollback recovery.
  • Plan-first, small-diff execution: Encourages planning before acting and making minimal diffs to reduce regression risk.
  • Concise, factual completion reporting: Requires exact changed files and commands, and forbids test-result overclaims.

Quick Start

Ask your agent to follow the omk-global-rules when coordinating a coding workflow so it avoids repeated instructions, uses subagents or relevant skills as needed, and finishes with an exact Changed/Files/Commands/Result/Risk report.

Frequently Asked Questions about omk-global-rules

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I enforce consistent agent coordination during multi-step coding workflows?

Multi-step coding workflows are standardized by enforcing global operating rules for plan-first execution, evidence-gated context transitions, and concise change reporting. This prevents inconsistent agent behavior across complex coordination tasks.

What is evidence-gated execution and how does it manage context transitions?

Evidence-gated execution requires verifiable checkpoints before risky context changes occur during agent coordination. It uses checkpoint preferences to validate progress and enable rollback recovery, ensuring safe transitions between workflow steps.

How do I get exact file change reporting from coding agents without overclaiming tests?

Coding agents produce concise, factual completion reports by listing exact changed files and executed commands. The global rules explicitly forbid test-result overclaims, ensuring accurate and transparent change reporting.

Does this plan-first approach work with MCP tools and project memory?

The plan-first approach supports safe use of MCP tools and project or session memory. It standardizes how agents interact with these components while maintaining evidence-gated verification and minimal diffs.

Why should I use subagents for complex coding tasks instead of a single agent?

Subagents reduce regression risk in complex coding tasks by enabling plan-first execution and small, isolated diffs. The global rules prefer delegating specialized work to subagents to avoid repeated instructions and maintain workflow consistency.