Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Changelog Principle

    2

    Write clean, human-centric CHANGELOGs from net diffs instead of noisy commit logs. Automatically drops internal refactors, folds unreleased polish into final features, and strictly follows Keep a Changelog.

    Free

    Security scanned
    changelog-principle

    changelog-principle

    Example session with this skill installed

    We are about to tag v1.3.0. Look at recent git commits and update the Unreleased section of CHANGELOG.md with our changes from this session.

    • Read your context and instructions
    • Compiled the changelog-principle

    I have inspected the changes against the previous release tag (v1.2.0).

    Following the skill

    changelog-principle, I derived the entries from the

    net diff rather than transcribing raw git commits. Although the commit log contains 7 commits (including internal refactors, styling iterations, and test fixes), only the user-observable outcomes are recorded.

    Net Diff Audit & Noise Filtering

    Dropped: refactor: extract auth provider into hook, test: add magic link unit tests, chore: bump dependencies (internal churn with no observable behavior change).

    Folded in: Iterative button styling tweaks landed in commit c3a8f1 were consolidated directly into the main authentication entry.

    Here is the updated CHANGELOG.md:

    
    # Changelog
    All notable changes to this project will be documented in this file.
    
    The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
    and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
    
    ## [Unreleased]
    
    ### Added
    
    - Magic link authentication, allowing users to sign in via a one-time link sent to their email.
    - Animated loading indicator on the login screen while the authentication request is in progress.
    
    ### Fixed
    
    - The submit button on the login form no longer remains disabled if an email delivery fails.
    

    Pre-save Checklist

    1. Every entry corresponds to an observable difference in the net diff.
    2. Internal refactors, tests, and temporary intermediate states were omitted.
    3. Reworked UI tweaks were edited in-place into the feature entry.
    4. Language is plain and direct, naming the exact outcome without filler praise.

    Next steps:
    Ready to commit. When tagging v1.3.0, rename [Unreleased] to [1.3.0] - 2026-11-10.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Stop turning your CHANGELOG into a developer diary.

    When AI coding agents update release notes, their default reflex is to read git log and paraphrase commit subjects. The result is noise: fix typo, wip, half-baked intermediate states, and internal refactors that end users never see.

    changelog-principle enforces one fundamental rule:

    "What will a user or developer notice that differs from the last release?"
    If the answer is nothing, write no entry.


    Key Capabilities

    • Net Diff as Source of Truth: Generates entries exclusively from the net diff against the last released state (git diff <tag>..HEAD), ignoring commit clutter.
    • Noise Elimination: Automatically filters out refactors, helper extractions, formatting, and dev-server tweaks.
    • Rework & Polish Folding: Tweaks to unreleased features are folded directly into the original feature entry—no phantom Fixed or Changed lines.
    • Keep a Changelog 1.1.0 Compliant: Cleanly categorizes into Added, Changed, Deprecated, Removed, Fixed, and Security.
    • Pre-save Quality Checklist: Verifies that every entry maps to an observable capability and eliminates all duplicate or phantom claims.

    Universal Compatibility

    Works out of the box with any AI coding agent:

    • Skills CLI: npx skills add NBDatsuya/changelog-principle -g -y
    • Native Agents: Direct drop-in for Claude Code, Cursor Rules (.mdc), Windsurf Rules, and Antigravity.

    Open Source & Upstream Repository

    This project is free and community-driven.

    Requires access to git diffs or commit logs. Does not handle automated versioning or deployment hooks. Best suited for projects following Keep a Changelog conventions.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 7 days ago

    • Free with an Agensi account

    Requires

    Node.js
    Git

    Permissions

    Terminal / Shell
    Read Files
    Write Files
    Listed7 days ago

    What's inside

    Frequently Asked Questions

    Instead of paraphrasing commit subjects—which inherit all the noise of WIPs and temporary experiments—it uses the net diff against your last release as the single source of truth. It logs only the final, observable outcome for the user and completely drops internal refactors and intermediate churn.

    This skill is designed to work with any AI agent capable of processing file diffs or commit histories and is compatible with projects following the "Keep a Changelog" standard.

    No, the skill focuses on content curation; it does not perform automated git tagging, version bumping, or repository writes, ensuring you maintain final control over your release process.

    The skill intelligently merges multiple experimental or transitional commits into a single, cohesive entry that describes the final net result for the user. It ignores the commit sequence and collapses intermediate states to the net result. If a feature went from A to B to C before release, only C is recorded. All polish, bugfixes, and refactors applied to unreleased code are folded directly into the original feature entry.

    Yes, this skill is 100% free and open-source under CC BY 4.0. You receive the complete SKILL.md specification, pre-save quality checklists, multi-language documentation, and direct compatibility with standard agent skill managers.

    New in Technical Documentation