Can it detect contradictions in my documentation?

Yes, it cross-references recent code changes and git logs against existing markdown documentation to flag contradictions that might otherwise confuse an AI agent.

What files does this skill create?

It creates or updates an AGENT_CHANGELOG.md file in your project's root directory, which serves as a specialized context layer for AI models.

How does this differ from a standard git changelog?

While standard changelogs focus on user-facing features, this skill focuses on architectural decisions, deprecated patterns, and internal rationale specifically to help AI agents code more accurately.

When should I run the Agent Changelog skill?

It is best to run this skill after major architectural changes, migrations, or when you notice that your AI assistant is suggesting outdated patterns or referencing stale docs.

Agent Context Changelog

Name: Agent Context Changelog
Author: petekp

bypetekp

0•

学习与文档

Generates agent-optimized changelogs to help AI coding assistants understand architectural evolution and avoid deprecated patterns.

The Agent Changelog skill automates the creation of an AGENT_CHANGELOG.md file, a specialized document designed specifically for AI coding agents to grasp the history and current state of a project. By cross-referencing Git logs, internal planning documents, and code signals like deprecation tags, it identifies architectural shifts and flags 'stale' documentation that might otherwise mislead an LLM. This ensures that AI agents have the correct context to make decisions aligned with current project standards, making it invaluable for onboarding assistants to legacy codebases or maintaining alignment after major refactors.

主要功能

01Generates a structured, machine-readable timeline optimized for LLM context windows.

02Automatically identifies and flags stale documentation contradicting current code reality.

03Cross-references git history with internal planning documents and documentation.

04Highlights deprecated patterns to prevent AI agents from suggesting legacy code.

050 GitHub stars

06Provides architectural rationale for key project milestones and decision history.

使用场景

01Onboarding an AI agent to an unfamiliar or legacy codebase for the first time.

02Synchronizing documentation with code reality after major architectural migrations or refactors.

03Establishing a 'source of truth' for project trajectory during rapid development phases.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add petekp/agent-skills agent-changelog

For use in Claude.ai and ChatGPT

Download Skill