Git Changelog Generator

Generate structured changelogs from git history using conventional commits, with support for multiple formats, AI-enhanced descriptions, and customizable ran...

Security Scan

The skill's code and instructions are consistent with a changelog generator that reads local git history and produce formatted output; it does not request credentials, install code, or perform unexpected network access.

Purpose & Capability

Name/description, SKILL.md usage examples, and the included Python script all align: the tool runs git log on a repository, parses conventional commits, groups and formats them. There are no unrelated environment variables, binaries, or install steps requested.

Instruction Scope

SKILL.md instructs running the bundled script and describes AI 'enhancements' (rewriting messages, cross-referencing issue trackers). The script itself performs only local git operations and formatting. The AI-enhancement language is high-level and gives the agent discretion to enrich output, but the skill does not include code or declared credentials to perform networked lookups; this is a mild vagueness rather than a mismatch.

Install Mechanism

No install specification is provided (instruction-only + included script). Nothing is downloaded or written to disk by an installer; risk from install mechanisms is minimal.

Credentials

The skill declares no required environment variables or credentials and the script does not read environment secrets. It only invokes git against a configurable repository path (default: current directory). No disproportionate credential access is requested.

Persistence & Privilege

always is false and the skill does not modify agent/global configuration or other skills. It runs on demand and writes output only to stdout or a user-specified file.

Assessment

This skill is coherent and low-risk: it runs local git commands to build changelogs and requires no credentials. Before running, review the script if you want to be sure it will only read the intended repository path (default: current directory). If you expect the AI to enrich changelogs with issue data or external context, be aware that the skill does not include network code or credentials—any such lookups would come from the agent itself or other skills, so only allow those if you trust the agent's configured integrations.

Generate structured changelogs from git history. Supports conventional commits, semantic versioning, and multiple output formats. Use when preparing releases, writing release notes, or documenting project history.

Usage

# Generate changelog for latest unreleased changes
python3 scripts/generate_changelog.py

# Generate changelog between two tags
python3 scripts/generate_changelog.py --from v1.2.0 --to v1.3.0

# Generate for last N commits
python3 scripts/generate_changelog.py --last 20

# Generate since a date
python3 scripts/generate_changelog.py --since 2026-04-01

Output Formats

# Markdown (default)
python3 scripts/generate_changelog.py --format markdown

# Keep a Changelog format (keepachangelog.com)
python3 scripts/generate_changelog.py --format keepachangelog

# GitHub Release format
python3 scripts/generate_changelog.py --format github-release

# JSON (for programmatic use)
python3 scripts/generate_changelog.py --format json

How It Works

1. Collect — reads git log between specified ranges
2. Parse — extracts conventional commit types (feat, fix, refactor, docs, test, chore, perf, ci)
3. Categorize — groups changes by type with human-readable headers
4. Enrich — adds PR links, issue references, author attribution, breaking change warnings
5. Format — outputs in the requested format

Conventional Commit Support

Parses standard prefixes:

• feat: → Features
• fix: → Bug Fixes
• refactor: → Code Refactoring
• docs: → Documentation
• test: → Tests
• perf: → Performance
• ci: → CI/CD
• chore: → Maintenance
• BREAKING CHANGE: → Breaking Changes (highlighted)

Non-conventional commits are categorized as "Other Changes" with AI-assisted categorization.

Parameters

Parameter	Description	Default
--from	Start tag/commit	Last tag
--to	End tag/commit	HEAD
--last	Last N commits	All since last tag
--since	Start date (YYYY-MM-DD)	None
--format	Output format	markdown
--output	Write to file	stdout
--repo	Repository path	Current directory
--include-authors	Show commit authors	false
--include-hashes	Show commit hashes	false
--group-by	Group by type or scope	type

AI Enhancement

When used as an agent skill, the AI can:

• Rewrite terse commit messages into human-readable descriptions
• Identify the most impactful changes and highlight them
• Generate a summary paragraph for release announcements
• Detect breaking changes even without conventional commit markers
• Cross-reference with issue trackers for richer context