When is the best time to run the validate-counts skill?

It is best used as a final check before creating a release, after bulk-adding components, or whenever the Project Overview in your CLAUDE.md looks outdated.

Does this tool modify my files automatically?

No, it is a validation-only tool that identifies discrepancies and reports them with precise locations so you can review and fix them manually or via build scripts.

Why is count validation important in AI development?

It prevents 'documentation drift' where your project overview or manifests claim to have a different number of skills or agents than what actually exists in the source code, which can confuse AI models and developers.

Can I use this for non-OrchestKit projects?

While specifically designed for OrchestKit file structures, the validation logic can be adapted for any project requiring consistency between source directories and configuration manifests.

What specific files does the validator check?

It compares the physical /src directory against CLAUDE.md, manifests/ork.json, and internal hook configuration files to ensure every component is accounted for.

Project Integrity Validator

Name: Project Integrity Validator
Author: yonatangross

byyonatangross

•

128

•

開発者ツール

Ensures structural consistency by validating hook, skill, and agent counts across project manifests, documentation, and source code.

The validate-counts skill is an essential utility for maintaining project integrity within OrchestKit environments by ensuring that the number of hooks, skills, and agents remains consistent across all authoritative sources. By cross-referencing the physical filesystem against CLAUDE.md, manifests, and internal JSON configurations, it automatically detects and flags documentation drift with precise file and field references. This automation is critical for large-scale AI development projects where manual counting is error-prone, providing a reliable gate for releases and general project maintenance to ensure the 'Project Overview' always reflects reality.

主な機能

01Automated cross-referencing between source directories and metadata files

02128 GitHub stars

03Pre-release validation gate integration

04Real-time drift detection for hooks, skills, and agents

05Fast execution via optimized bash scripts for raw component counting

06Detailed comparison tables with precise file and field references

ユースケース

01Resolving inconsistencies in the CLAUDE.md Project Overview section

02Auditing project structure after adding or removing multiple components

03Pre-release verification to ensure documentation matches the actual codebase

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add yonatangross/orchestkit validate-counts

For use in Claude.ai and ChatGPT

Download Skill