Why does AI need different documentation than humans?

Humans can often infer missing steps or context; AI agents require literal, step-by-step instructions with explicit file paths and defined success criteria to avoid errors.

What is a project health audit for AI agents?

It is a specialized evaluation that ensures your project's documentation, paths, and workflows are explicit enough for an AI to follow literally without needing human clarification.

Can this skill help with documentation debt?

Yes, it specifically targets 'AI documentation debt' by flagging outdated workflows or vague instructions that would typically cause an AI agent to stall or fail.

When should I use the handoff-checker?

You should use the handoff-checker proactively before ending a Claude Code session to ensure all progress, blockers, and next steps are documented for the next session.

How does the context-auditor improve AI sessions?

The context-auditor identifies ambiguous references and implicit knowledge in your docs, providing a score and specific recommendations to make them more 'AI-readable'.

Project Health & AI Readiness

Name: Project Health & AI Readiness
Author: jezweb

byjezweb

•

245

•

协作工具

Audits project documentation and workflows to ensure they are fully executable and understandable by AI agents.

The Project Health skill optimizes how developers and AI agents collaborate by auditing repository documentation for 'agent-readability.' It provides a specialized suite of agents—context-auditor, workflow-validator, and handoff-checker—that evaluate whether project resources are explicit, literal, and structured for seamless session continuity. By identifying documentation debt and ambiguous instructions, this skill ensures that future Claude Code sessions can resume work effectively, follow workflows without failure, and maintain context across complex development cycles.

主要功能

01Context Auditor to measure AI-readability and resolve documentation ambiguity.

02Standardized templates for creating AI-readable documentation and handoff checklists.

03Workflow Validator to ensure process steps are discrete, ordered, and literally executable.

04Automated scoring metrics (0-100) to quantify project health from an AI perspective.

05245 GitHub stars

06Handoff Checker for validating session continuity and context preservation via SESSION.md.

使用场景

01Preparing a project for a clean handoff between different Claude Code sessions.

02Onboarding a new AI agent or human team member to a complex codebase.

03Auditing workflows after major refactors to ensure they remain executable by AI.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jezweb/claude-skills project-health

For use in Claude.ai and ChatGPT

Download Skill