
release-note-generation
Toolkit for generating PowerToys stable or preview release notes from GitHub milestones, commit ranges, or Azure DevOps release-candidate builds. Use when asked to create release notes, summarize milestone PRs, generate changelog, prepare a draft preview release, calculate PR deltas across main and stable, update release documentation, manage PR milestones, or prepare and validate release assets.
- Installs
- 117
- Last 24h
- —
- Trending rank
- —
- Security
- Not audited
Install
Works with every agent the skills CLI supports.
npx skills add https://github.com/microsoft/powertoys --skill release-note-generation- 01
Install
Run the command in your project. The CLI asks which agents to install for.
- 02
Describe the task
Ask your agent for something this skill covers. It loads release-note-generation automatically when relevant.
- 03
Review the result
The agent follows the SKILL.md steps, scripts and references bundled with the skill.
SKILL.md
The instructions your agent reads when the skill is triggered.
Release Note Generation Skill
Generate professional PowerToys release notes by collecting merged PRs, summarizing each PR with the local CLI agent, grouping by label, and producing user-facing summaries. Stable and preview releases share the same PR metadata, attribution, grouping, and formatting rules.
Output Directory
All generated artifacts are placed under Generated Files/ReleaseNotes/ at the repository root (gitignored).
Generated Files/ReleaseNotes/
├── milestone_prs.json # Raw PR data from GitHub
├── sorted_prs.csv # Sorted PR list with Copilot summaries
├── prs_with_milestone.csv # Milestone assignment tracking
├── grouped_csv/ # PRs grouped by label (one CSV per label)
├── grouped_md/ # Generated markdown summaries per label
└── v{VERSION}-release-notes.md # Final consolidated release notes
Preview-release runs use an isolated subdirectory:
Generated Files/ReleaseNotes/preview-<buildId>/
├── release-context.json
├── delta-commits.json
├── delta-prs.json
├── removed-prs.json
├── unattributed-commits.json
├── MemberList.md
├── milestone_prs.json
├── sorted_prs.csv
├── release-notes.md
├── hashes.md
├── release-manifest.json # Local audit artifact; never uploaded
├── assets-manifest.json # Local asset inventory; never uploaded
└── final-review.md
When to Use This Skill
- Generate release notes for a milestone
- Summarize PRs merged in a release
- Generate per-PR review summaries locally for release-notes copy
- Assign milestones to PRs missing them
- Collect PRs between two commits/tags
- Update README.md for a new version
- Prepare GitHub release assets (download installers/symbols + compute hashes)
- Prepare a complete draft preview release from an ADO build URL or build ID
- Compare preview contents across
mainandstablebranch transitions
Prerequisites
- GitHub CLI (
gh) installed and authenticated — The collection script usesgh pr viewandgh api graphqlto fetch PR metadata and co-author information. Rungh auth statusto verify; if not logged in, rungh auth loginfirst. See Step 1.0.0 for details. - MCP Server: github-mcp-server installed (used to fetch PR diffs/files for the local-agent review step)
- For preview releases and prepare-release-assets.ps1: Azure CLI authenticated against the Microsoft tenant (
az login) with theazure-devopsextension; access to themicrosoft/DartADO project
Required Variables
For a stable release, confirm {{ReleaseVersion}} with the user before starting.
For a preview release, do not request a version: derive it from the candidate ADO build.
| Variable | Description | Example |
|---|---|---|
{{ReleaseVersion}} | Target release version | 0.98 |
Preview mode instead requires one ADO build URL or numeric build ID. It derives the version, source commit, branch, and previous release without asking the user.
Scenario routing
Read the scenario index, then follow only the selected scenario:
- Stable release for milestone- or version-based release notes.
- Preview release for an autonomous ADO-build-to-draft workflow.
Workflow Overview
┌────────────────────────────────┐
│ 1.0 Verify gh auth + MemberList │
└────────────────────────────────┘
↓
┌────────────────────────────────┐
│ 1.1 Collect PRs (stable range) │
└────────────────────────────────┘
↓
┌────────────────────────────────┐
│ 1.2 Assign Milestones │
└────────────────────────────────┘
↓
┌────────────────────────────────┐
│ 2.1–2.4 Label PRs (auto+human) │
└────────────────────────────────┘
↓
┌────────────────────────────────┐
│ 3.1 Local-agent PR summaries │
│ (writes CopilotSummary) │
└────────────────────────────────┘
↓
┌────────────────────────────────┐
│ 3.2 (Optional) Refresh PR data │
└────────────────────────────────┘
↓
┌────────────────────────────────┐
│ 3.3 Group by label │
│ (grouped_csv) │
└────────────────────────────────┘
↓
┌────────────────────────────────┐
│ 4.1 Summarize (grouped_md) │
└────────────────────────────────┘
↓
┌────────────────────────────────┐
│ 4.2 Final notes (v{VERSION}.md) │
└────────────────────────────────┘
| Step | Action | Details |
|---|---|---|
| 1.0 | Verify prerequisites | gh auth status must pass; generate MemberList.md |
| 1.1 | Collect PRs | From previous release tag on stable branch → sorted_prs.csv |
| 1.2 | Assign Milestones | Ensure all PRs have correct milestone |
| 2.1–2.4 | Label PRs | Auto-suggest + human label low-confidence |
| 3.1–3.3 | Reviews & Grouping | Local agent summarizes each PR diff into CopilotSummary → (optional refresh) → group by label |
| 4.1–4.2 | Summaries & Final | Generate grouped summaries, then consolidate |
Detailed workflow docs
Do not read all steps at once—only read the step you are executing.
- Step 1: Collection & Milestones
- Step 2: Labeling PRs
- Step 3: Reviews & Grouping
- Step 4: Summarization
Available Scripts
| Script | Purpose |
|---|---|
| dump-prs-since-commit.ps1 | Fetch PRs between commits/tags |
| group-prs-by-label.ps1 | Group PRs into CSVs |
| collect-or-apply-milestones.ps1 | Assign milestones |
| diff_prs.ps1 | Incremental PR diff |
| prepare-release-assets.ps1 | Download installers + symbols from an ADO build, compute SHA256, emit the "Installer Hashes" markdown table for the GitHub release page |
| get-release-build-metadata.ps1 | Resolve and validate the candidate build identity, version, channel, intent, and source commit |
| get-previous-published-release.ps1 | Select the latest published stable or preview release that predates the candidate queue time |
| get-preview-release-delta.ps1 | Calculate semantic added/removed PRs between exact release commits |
| collect-pr-metadata.ps1 | Normalize GitHub metadata for an explicit set of PR numbers |
| new-preview-release-manifest.ps1 | Create the auditable build, baseline, and semantic-delta release manifest |
| upsert-draft-preview-release.ps1 | Create or update a draft prerelease without exposing a publish operation |
| verify-draft-preview-release.ps1 | Verify draft flags, target commit, managed body, and uploaded assets |
References
- Sample Output - Example summary formatting
- Detailed Instructions - Legacy full documentation
Conventions
- Terminal usage: Disabled by default; only run scripts when user explicitly requests
- Preview automation: An explicit request to prepare a preview release, or invocation by the Prepare Preview Release agent, authorizes the canonical preview scripts
- Preview manifests: Keep
release-manifest.jsonandassets-manifest.jsonin the local audit package; do not upload either file as a GitHub release asset - Preview note layout: Place the
Installer Hashessection immediately after the title and short public introduction, beforeHighlightsand all change sections - Batch generation: Generate ALL grouped_md files in one pass, then human reviews
- PR order: Preserve order from
sorted_prs.csvin all outputs - Label filtering: Keeps
Product-*,Area-*,GitHub*,*Plugin,Issue-*
Troubleshooting
| Issue | Solution |
|---|---|
gh command not found | Install GitHub CLI and add to PATH |
| No PRs returned | Verify milestone title matches exactly |
Empty CopilotSummary for many PRs | Run Step 3.1 (local-agent summaries). Do not use mcp_github_request_copilot_review from a CLI/coding agent — the GitHub API rejects bot-initiated review requests, so the column will stay empty. |
| Many unlabeled PRs | Return to labeling step before grouping |
prepare-release-assets.ps1 fails with "Failed to acquire ADO access token" | Run az login and ensure you have access to the microsoft/Dart ADO project |
Candidate has no release-metadata.json | The metadata resolver uses pipeline-log fallback; ambiguous or conflicting values stop the run |
| First preview after switching branches has unexpected changes | Review removed-prs.json and unattributed-commits.json; see preview delta resolution |
FAQ
What does the release-note-generation skill do?
Toolkit for generating PowerToys stable or preview release notes from GitHub milestones, commit ranges, or Azure DevOps release-candidate builds. Use when asked to create release notes, summarize milestone PRs, generate changelog, prepare a draft preview release, calculate PR deltas across main and stable, update release documentation, manage PR milestones, or prepare and validate release assets.
How do I install release-note-generation?
Run npx skills add https://github.com/microsoft/powertoys --skill release-note-generation in your project. The skills CLI detects the agents you use and copies the skill into the right folder. For Claude Code you can also place the skill folder at ~/.claude/skills/release-note-generation/ (personal) or .claude/skills/release-note-generation/ (project).
Which AI agents can use release-note-generation?
Any agent that supports the open Agent Skills format (a folder with a SKILL.md file), including Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Windsurf, Cline, OpenCode and more.
Is release-note-generation safe to install?
No partner security audit is available yet. Skills can run commands and scripts, so read the SKILL.md and any bundled files before installing.
Who maintains release-note-generation?
It is published by microsoft in microsoft/powertoys, and it is an official skill from the maker of the technology it covers. It has 117 recorded installs.
Popular code quality & workflow skills
grill-me
mattpocock/skills
A relentless interview to sharpen a plan or design.
improve-codebase-architecture
mattpocock/skills
Scan a codebase for deepening opportunities, present them as a visual HTML report, then grill through whichever one you pick.
handoff
mattpocock/skills
Compact the current conversation into a handoff document for another agent to pick up.
domain-modeling
mattpocock/skills
Build and sharpen a project's domain model. Use when discussing codebase terminology, writing or editing a GLOSSARY.md, or recording or editing an ADR.
codebase-design
mattpocock/skills
Shared vocabulary for designing deep modules. Use when the user wants to design or improve a module's interface, find deepening opportunities, decide where a seam goes, make code more testable or AI-navigable, or when another skill needs the deep-module vocabulary.
diagnosing-bugs
mattpocock/skills
Diagnosis loop for hard bugs and performance regressions. Use when the user says "diagnose"/"debug this", or reports something broken/throwing/failing/slow.