Changelog & Release Notes Generator
Auto-generates changelogs from commits, PRs, and tickets
View the interactive project page →
Changelog & Release Notes Generator
AI-powered changelog and release notes automation that generates human-quality summaries from unstructured commit messages, PR descriptions, and issue tickets — eliminating the "what changed" documentation burden.
The Problem
Release note generation is trapped in a specification straitjacket:
- Rule-based tools require Conventional Commits — ~90% of projects don't use them
- Detection-only tools stop at diagnosis — semantic-release and git-cliff surface what changed; none generate what it means to users
- The developer-to-user gap is unsolved — every tool produces internal
CHANGELOG.mdbut no tool generates user-facing "What's New" announcements from the same source - Issue tracker context is ignored — Jira/Linear tickets have business context that git commits lack; no tool synthesizes this
Current state: developers write changelogs manually (tedious, error-prone) or rely on auto-generated lists of commit subjects (low quality, developer-facing only).
What This Does
Free-Text Commit Classification (AI-Native)
- Reads unstructured commit messages — no Conventional Commits required
- LLM infers type (feature/fix/breaking/internal) and scope from commit message, diff summary, and PR description
- Works out-of-the-box for 90% of projects — the gap that rule-based tools fail on
- Handles inconsistent conventions — mixed Conventional Commits + free-text in the same repo
Dual-Audience Output
- Generates both developer CHANGELOG and user-facing "What's New" from the same source
- Developer output: technical details (API changes, deprecated methods, internal refactors)
- User output: benefit statements (features added, problems fixed, performance improved)
- Internal refactors filtered out of user-facing notes automatically
- No open-source tool does this — gap between Beamer/Headway (SaaS, user-facing only) and semantic-release/git-cliff (OSS, internal only)
Issue Tracker Integration
- Fetches linked Jira/Linear/GitHub Issues for context
- Incorporates business-level descriptions — not just code-level reasoning
- Produces better release notes by understanding why the change matters, not just what changed
- Attribute changes to business priorities — which features delivered for which customer requests
Role-Aware Personalization
- Different release notes for different roles — inspired by SmartNote research (2025)
- Admin variant: security fixes, performance improvements, operational notes
- User variant: new features, problem solutions, benefit statements
- Developer variant: API changes, deprecations, internal refactors
Key Differentiators
| Feature | This Platform | semantic-release | git-cliff | release-please | WhatShipped | Beamer |
|---|---|---|---|---|---|---|
| Free-text commits | ✓ (LLM-based) | Requires Conventional | Requires structure | Requires Conventional | ✓ (Cloud) | — |
| Dual-audience output | ✓ | — | — | — | — | ✓ SaaS only |
| Issue tracker integration | ✓ (Jira/Linear) | — | — | — | — | — |
| Role-aware variants | ✓ | — | — | — | — | — |
| Open source | ✓ | ✓ (MIT) | ✓ (MIT) | ✓ (Apache) | — (SaaS) | — (SaaS) |
| CI/CD integration | ✓ | ✓ | ✓ | ✓ | — | — |
Market & Opportunity
- Market size: $2.11B (2024) → $4.5B (2035) at 7.1% CAGR (release management subset)
- Adoption barrier: ~90% of projects don't use Conventional Commits — rule-based tools are useless for them
- Buyers: Development teams (all sizes), DevOps engineers, product managers, technical writers
- Open-source gap: No OSS tool generates both developer and user-facing notes from the same source
Research Foundation
- ~90% of projects don't enforce Conventional Commits — research finding from ICSE 2025 study
- 52 documented classification challenges in Conventional Commits (ICSE 2025)
- 14% of commits are empty; 66% have minimal descriptions (arXiv:2202.02974, 2022)
- Commit message quality correlates with software defect proneness (ICSE 2023)
- LLM-based commit classification outperforms rule-based approaches by 12 F1 points (Fäerber et al., COMPSAC 2023)
Quick Start
# Generate changelog from git history
changelog-gen --since=v1.0.0 --until=HEAD --output=CHANGELOG.md
# Generate dual-audience output
changelog-gen --since=v1.0.0 \
--developer-output=CHANGELOG-DEV.md \
--user-output=WHATSNEW.md
# Integrate with issue tracker
changelog-gen --jira-url=https://jira.company.com \
--jira-project=MYAPP \
--include-context
# In CI/CD pipeline
changelog-gen --auto --commit-to=main --create-release
Target Users
- Development Teams (all sizes) — eliminate manual changelog writing
- DevOps/Release Engineers — automate release note generation in CI/CD
- Product Managers — generate user-facing "What's New" announcements
- Technical Writers — auto-generate draft release notes for refinement
- Open-source Projects — good community communication with minimal effort
Related Standards
- Conventional Commits Specification (v1.0.0) — machine-parseable commit format
- Semantic Versioning (SemVer v2.0.0) — version signaling aligned with changes
- Keep a Changelog (v1.1.0) — human-facing changelog format standard
- Git Tagging Conventions — release tag naming and management
Built on research from SmartNote (arXiv:2505.17977, 2025), ICSE 2025 Conventional Commits classification study, and production learnings from semantic-release, release-please, and WhatShipped. Read the full research | Feature roadmap