REA의 분석 처리와 캐시 설계를 검토합니다. 결과의 정확성과 근거 접근성을 유지하면서 처리 시간, CPU, 메모리, 에이전트가 다루는 정보량을 개선합니다.
- 분석 처리가 느린 원인 조사
- 캐시 설계 검토
- CPU와 메모리 부하 절감
REA의 Git 이력을 특정 버전으로 고정해 변경 이력과 릴리스 노트를 정리합니다. 기여자 감사 표기를 검증하고 Release Please의 공개 본문과 일치하는지 확인합니다.
Prepare or rewrite REA release changelogs and GitHub release notes from pinned Git history, with verified contributor thanks and Release Please synchronization.
rea-changelog-update는 REA의 Git 이력에서 대상 릴리스의 변경을 찾아 CHANGELOG.md와 GitHub 릴리스 노트를 작성하거나 수정합니다. 기준 태그와 대상 커밋을 고정하고 PR, 직접 커밋, 공동 작성자, 수정된 이슈의 보고자를 확인합니다. 사용자에게 미치는 영향에 따라 내용을 묶고 주요 내용, 변경, 수정, 호환성을 깨는 변경으로 정리합니다.
자동 생성된 초안을 읽기 쉽게 다듬거나 기여자 감사 표기를 누락과 중복 없이 정리할 때 유용합니다. 공개 본문에서 제외한 내부 작업도 기록에 남겨 나중에 서술의 근거를 확인할 수 있습니다. Release Please로 릴리스를 준비할 때는 변경 이력과 PR 본문의 일치 여부도 검증합니다.
전체 Git 이력과 태그, 관련 GitHub 정보, 기존 생성 및 검증 스크립트를 사용합니다. 확인되지 않은 계정명은 추측하지 않습니다. 진행 중인 PR을 수정하려면 권한이 필요하며, 이 스킬 자체는 병합, 태그 지정, 게시를 허용하지 않습니다.
이 소개문은 공개된 SKILL.md를 바탕으로 AI(OpenAI Codex)가 작성했습니다. 정확한 사양은 아래 원문을 확인하세요.
AI가 SKILL.md의 내용만을 근거로 4가지 관점에서 판정했습니다. 스타 수나 인기도는 판정에 넣지 않았습니다.
총점은 각 항목을 A=25점, B=15점, C=5점으로 더한 값입니다. 2026년 10월 19일 기준으로 AI(OpenAI Codex)가 판정했습니다. 참고용이므로 설치 전에 원문도 확인해 주세요.
설치하기 전에 에이전트에게 전달되는 지시 내용을 확인할 수 있습니다.
Use this for release-note preparation and contributor attribution. Read docs/releasing.md from the repository root for publication authority and the existing checkpoint checks. This skill prepares notes; it does not authorize merging, tagging, or publishing.
Fetch complete history and tags. Select the published baseline tag from
.release-please-manifest.json on the release PR's base branch, and pin that
branch's exact SHA as the editorial target. REA tags are rea-agents-VERSION.
Use the proposed version, including its prerelease suffix; do not change the
always-bump-minor strategy to imitate OpenClaw's calendar versions.
node scripts/release-notes.mjs inventory \
--repository morluto/rea --base rea-agents-PREVIOUS \
--target FULL_BASE_BRANCH_SHA --version VERSION \
--manifest /tmp/rea-VERSION.contributions.json
node scripts/release-notes.mjs record \
--manifest /tmp/rea-VERSION.contributions.json \
--summary \
--output /tmp/rea-VERSION.record.md
Read the inventory, not just the existing bot draft. It includes side-branch commits, GitHub-associated PRs whose merge commits are in the range, commit authors/co-authors, and confirmed closing issues with their reporters. GitHub connections are paginated. Network errors stop generation; rerun into a new output file after recovery. Existing output files are never overwritten.
Unresolved commit identities are retained as names/emails with user: null;
never guess a handle or thank the merger as a substitute. The report preserves
unknowns. This is a reviewed metadata snapshot, not independent proof that a
human authored a patch. For unattributed identities, unusual merge strategies,
backports, reversions, or references that GitHub did not associate, inspect
the source PRs and diffs and report the limitation. The current collector uses
confirmed PR-closing relationships for issue-reporter credit; unrelated issue
mentions are not automatically credited. Do not claim that merely inventoried
work still ships after a revert.
Rewrite only the newest ## [VERSION] section in CHANGELOG.md, preserving its
version/date/link heading. Use single-line Markdown list items under:
### Highlights: a few release-defining user outcomes; no quota for small releases.### Changes: capabilities and meaningful behavior changes.### Fixes: reliability, correctness, compatibility, and safety fixes.### ⚠ BREAKING CHANGES: retain when applicable, with migrations and original
PR/commit references required by the existing checkpoint validator.Group related work by analyst workflow or user impact. Explain the resulting behavior in short, concrete entries. Highlights summarize the release; avoid repeating their full descriptions in Changes. Internal cleanup, tests, CI, and routine documentation generally belong only in the provenance file. Read ambiguous PRs/diffs before making claims. Distinguish simulated-provider checks from real-provider verification.
Credit each represented contribution once in the public notes with its link
and Thanks @... for its verified human author, known co-authors, and confirmed
closing-issue reporters. Grouping preserves all those credits in one line.
When Highlights, detailed entries or migrations reference the same work,
put its thanks in one of those entries, not all of them. Include issue links.
Direct-commit bullets carry the commit link and known human credit.
Exclude GitHub bots and known agent accounts. REA's observed Anthropic, Cursor
and Codex co-author trailers resolve to @claude, @cursoragent and @codex
with GitHub type User; the credit policy excludes them while retaining their metadata.
At the REA maintainer's request, omit @morluto self-thanks while preserving
authorship in the inventory. Other human contributors remain eligible.
Use explicit GitHub links for cross-repository references to avoid ambiguous
#NNN numbers. Unknown handles stay unknown; they are not fabricated credit.
Append the generated record --summary provenance link verbatim at the end
of this release. The linked inventory accounts for every discovered in-range
PR and direct contribution, including internal work omitted from the prose.
Do not paste the full record output into public notes: it duplicates thanks
and obscures the user-facing changes. Full records remain available for review.
Do not hand-edit generated credits to make validation pass. Refresh the inventory
when product history changes; a prose-only correction can reuse the snapshot.
Before changing the live release PR, have authorization for that PR update. Local preparation and validation can proceed independently.
Add rea:release-notes-finalized to the selected open release PR before
editing its branch/body. Main proposal pushes respect this label. This is
our workflow guard, not Release Please's autorelease: snooze mechanism.
A proposal already running must finish before editing. Do not dispatch
manual preparation over finalized notes.
On that release branch, save the reviewed inventory as
docs/releases/VERSION.contributions.json. This is durable public release
provenance, not a temporary planning file. Do not add credentials or local
environment dumps. Keep the existing version artifacts aligned.
Run the existing checkpoint validator and the new checker:
node scripts/release-notes.mjs check \
--manifest docs/releases/VERSION.contributions.json
node scripts/release-notes.mjs render \
--manifest docs/releases/VERSION.contributions.json \
--output /tmp/rea-VERSION.notes.md
The checker supports inline PR bodies, including their surrounding metadata, up to GitHub's 65,536-character limit. It fails on larger notes rather than truncating credit; Release Please's overflow-link bodies are not integrated.
Update the release PR body with the rendered section while preserving its
Release Please header/footer and component/version structure. Use a body
file for gh pr edit --body-file; changing only CHANGELOG.md does not
change Release Please's release body. Push only authorized changes.
Freshly fetch the PR body to a file and run check --pr-body FILE. Review
the complete diff and wait for current CI. Publication checks the body
again before release creation. Avoid further body edits while publishing.
The finalization label freezes bot refreshes, not main. New product changes after the inventory target require refreshing the notes before publication. The workflow checks target ancestry and allows only release metadata paths after that target. It also requires the inventory target to equal the reviewed merge's first parent, so intervening main commits require a fresh inventory. Review diffs within the allowed release metadata paths too: the path guard does not prove that package dependency/script edits are version-only changes. To resume automatic proposals, remove the finalization label and let the bot regenerate; expect editorial changes to be replaced.
Existing releases without a contribution inventory retain their current publication path. Never retag or republish a package to correct release prose.
아직 리뷰가 없습니다. 사용해 본 소감을 남겨 주세요.
로그인 없이 쓸 수 있습니다. 사용 후기, 질문, 작동하지 않은 환경 등을 편하게 남겨 주세요.
아직 댓글이 없습니다. 첫 댓글을 남겨 보세요.
REA의 분석 처리와 캐시 설계를 검토합니다. 결과의 정확성과 근거 접근성을 유지하면서 처리 시간, CPU, 메모리, 에이전트가 다루는 정보량을 개선합니다.
REA의 CI와 테스트 모음을 조사하여 결함을 놓치는 판정과 테스트 데이터의 차이를 검증합니다. 필요한 검출 능력을 유지하면서 실측을 바탕으로 실행 비용도 검토합니다.
REA 조사 도구를 설계하거나 변경할 때 담당 범위, 입출력, CLI 및 MCP 호출 규칙을 정리합니다. 조사 근거의 의미와 검증 방법을 포함한 설계안을 작성합니다.
REA로 배포된 앱, 바이너리, 패키지를 분석해 기능과 버전별 차이를 조사합니다. 역컴파일과 실행 중 관찰을 활용해 근거, 추론, 미확인 사항을 구분합니다.