jackin-release-notes
Use only when the user explicitly requests this skill. Classifies every PR merged since the last tag into Keep a Changelog categories and writes the [Unreleased] section of CHANGELOG.md.
Works with
---
name: jackin-release-notes
description: Use only when the user explicitly requests this skill. Classifies every PR merged since the last tag into Keep a Changelog categories and writes the [Unreleased] section of CHANGELOG.md.
license: Apache-2.0
---
# jackin-release-notes
**Classify** every PR merged since the last release tag into a Keep a Changelog bucket, then write the `[Unreleased]` section of `CHANGELOG.md`. The bar is exhaustive — every merged PR accounted for, none dropped, none doubled.
## When to use
- Before a release, to populate the changelog.
- When the operator asks to preview what the next release would contain.
## When NOT to use
- Cutting the release itself → `jackin-release` (it runs this skill as its notes step).
## Process
1. **Find the boundary.** `LAST_TAG=$(git describe --tags --abbrev=0)`; if none, start from the initial commit (`git rev-list --max-parents=0 HEAD`). `TAG_DATE=$(git log -1 --format=%aI "$LAST_TAG")`.
*Done when* you hold the boundary commit and its date.
2. **Gather merged PRs.** `gh pr list --state merged --base main --search "merged:>$TAG_DATE" --json number,title,labels,mergedAt`.
*Done when* every PR merged after the tag is in the list.
3. **Classify each.** Sort by Conventional Commits prefix and label:
| Prefix / label | Bucket |
|---|---|
| `feat:` / `feature` | **Added** |
| `fix:` / `bug` | **Fixed** |
| `security:` | **Security** |
| `refactor:`, `chore:`, `docs:`, `ci:`, `build:` | **Changed** |
| `deprecate:` | **Deprecated** |
| `remove:` | **Removed** |
| no match | **Changed** (default) |
Strip the prefix from the title; capitalize the first word (`feat: add TUI launcher` → `Add TUI launcher`).
*Done when* every PR sits in exactly one bucket.
4. **Catch ungrouped commits.** `git log <last-tag>..HEAD --oneline --no-merges`; for each, `gh pr list --state merged --search "<sha>"`. Commits with no PR land under **Ungrouped commits**.
*Done when* every non-merge commit since the tag is matched to a PR or listed.
5. **Write `[Unreleased]`.** Keep a Changelog format: one section per non-empty bucket, each entry linking its PR (`[#N](…/pull/N)`). Replace the existing `[Unreleased]` block (between `## [Unreleased]` and the next `## [`) — ask before overwriting entries an operator already edited. Keep the `<!-- next-header -->` marker intact.
*Done when* `CHANGELOG.md` carries every merged PR under the right bucket, empty buckets omitted, and the marker survives.
6. **Present for review.** Show the section; let the operator re-classify or reword. Do not proceed to a release until approved.
## Idempotency
Safe to re-run: reads `CHANGELOG.md` fresh each time, asks before overwriting existing entries, fetches PR data live from GitHub.
## Common mistakes
- Dropping a PR that has no conventional-commit prefix — it defaults to **Changed**, never omitted.
- Listing an empty bucket ("no **Removed** entries" is noise — drop the section).
- Forgetting the PR link in each entry.
- Overwriting an operator-edited `[Unreleased]` without asking.More Writing & Documentation skills
paper-context-resolver
lllllllama/rigorpilot-skills
Rigor Paper Context helper for README-first deep learning repo reproduction. Use only when the README and repository files leave a narrow reproduction-critical gap and the task is to resolve a specific paper detail such as dataset split, preprocessing, evaluation protocol, checkpoint mapping, or runtime assumption from primary paper sources while recording conflicts. Do not use for general paper summary, repo scanning, environment setup, command execution, title-only paper lookup, or replacing README guidance by default.
repo-intake-and-plan
lllllllama/rigorpilot-skills
Rigor Intake helper for README-first deep learning repo reproduction. Use when the task is specifically to scan a repository, read the README and common project files, extract documented commands, classify inference, evaluation, and training candidates, and return the smallest trustworthy reproduction plan to the main orchestrator. Do not use for environment setup, asset download, command execution, final reporting, paper lookup, or end-to-end orchestration.
minimal-run-and-audit
lllllllama/rigorpilot-skills
Rigor Run skill for README-first deep learning repo reproduction. Use when the task is specifically to capture or normalize evidence from the selected smoke test or documented inference or evaluation command and write standardized `repro_outputs/` files, including patch notes when repository files changed. Do not use for training execution, initial repo intake, generic environment setup, paper lookup, target selection, hidden scientific-meaning changes, or end-to-end orchestration by itself.
