Prepare, launch, monitor, and summarize the real RFdiffusion to ProteinMPNN to Protenix antibody pipeline on a local or remote ScienceDiscovery Runner with sandboxed Ascend NPUs.
日本語の概要は準備中です。原文の説明を表示しています。
Use the gh CLI for GitHub issues and pull requests on openJiuwen-ai/sciencediscovery. Issue and pull request titles and bodies are English. Use when filing or editing an issue, opening or updating a pull request, or when an agent or a person asks how to talk to this repository on GitHub. For the test gates, fork push, and release label, continue in the create-github-pr skill. gitcode.com is a read-only mirror.
日本語の概要は準備中です。原文の説明を表示しています。
実行スコア 未検証
インストール方法を見るインストールする前に、エージェントに与えられる指示の中身を確認できます。
gh)Project-local skill for ScienceDiscovery.
New issues and pull requests go to
openJiuwen-ai/sciencediscovery
with gh. Do not open them in the browser, and do not use the gitcode
CLI for a new issue or pull request. GitCode
(openJiuwen/sciencediscovery) is a read-only mirror. The
gitcode skill is only for an existing mirror item that
still has to be read or commented on.
The full pull-request procedure — local UT/ST/E2E, pushing the branch to your own fork, the release label, reading Actions back — is create-github-pr. This page is the usage guide those steps assume.
Issue and pull request titles and bodies are English.
That includes the explanation under any template heading, the validation
numbers, and the E2E conclusion. Write Chinese only when the person who asked
explicitly wants Chinese. If a GitHub issue or pull request was filed in
Chinese by mistake, fix it with gh issue edit or gh pr edit before calling
it done.
Edit only items you created. On someone else's issue or pull request, add a comment; do not rewrite their title, body or comments. The one exception is adding yourself as an assignee to an issue you start implementing, under Issues.
gh auth status # retry once if it fails, then ask the operator
gh api user --jq .login # current login; do not hard-code a person; never print a token
Always pass --repo openJiuwen-ai/sciencediscovery so a checkout whose
origin is something else still hits this repository.
origin on a current clone of this repo is
openJiuwen-ai/sciencediscovery. Do not git push origin a task branch.
Push the branch to the operator's own fork, then open the pull request with
--head <login>:<branch>.
Search before creating. A duplicate title is not a new issue.
gh issue list --repo openJiuwen-ai/sciencediscovery --state open --limit 50 \
--search "keywords"
gh issue create --repo openJiuwen-ai/sciencediscovery \
--title "[Feature]: Short English summary" \
--body-file .tmp/issue.md
gh issue view <n> --repo openJiuwen-ai/sciencediscovery
gh issue edit <n> --repo openJiuwen-ai/sciencediscovery --body-file .tmp/issue.md
gh issue comment <n> --repo openJiuwen-ai/sciencediscovery --body-file .tmp/comment.md
Read the issue back after create or edit and confirm the number, the English title, and that the body is the text you wrote.
To put a screenshot in the body, reference a local image in the Markdown and
pass the same path to --attach. gh rewrites the reference to a
user-attachments URL. --attach needs write access on this repository. A
person can also drop the file into the issue on the website; that upload does
not require a git push, but agents should use --attach rather than asking
someone to paste by hand.
gh issue edit <n> --repo openJiuwen-ai/sciencediscovery \
--body-file .tmp/issue.md \
--attach '.tmp/screen.png#What the screen shows'
These steps apply when you start implementing an existing issue. Reading it, giving a brief answer, or creating a new issue alone does not claim the work.
Resolve the current gh login and inspect the issue's assignees:
login=$(gh api user --jq .login)
gh issue view <n> --repo openJiuwen-ai/sciencediscovery \
--json number,title,assignees
If you are already assigned, skip the assignment. Otherwise add yourself;
--add-assignee preserves everyone already assigned:
gh issue edit <n> --repo openJiuwen-ai/sciencediscovery --add-assignee @me
If @me is unavailable, use the resolved login instead; never hard-code a
person's login:
gh issue edit <n> --repo openJiuwen-ai/sciencediscovery --add-assignee "$login"
Read the issue back and check whether assignees contains your login:
gh issue view <n> --repo openJiuwen-ai/sciencediscovery \
--json number,assignees
If your login is missing, do not retry assignment or report it as assigned.
Still post the plan comment in step 4, say you are working on the issue,
and ask a maintainer to assign it to you. The "$login" fallback above only
covers @me being unavailable; it does not grant assignment permissions.
Post an English plan whether or not assignment persisted, describing what you intend to change and how completion will be judged. If other assignees are present, say that you are also working on it. Add a new English progress comment when a PR opens, work is blocked or the plan changes. Leave other people's titles, bodies and comments unchanged:
gh issue comment <n> --repo openJiuwen-ai/sciencediscovery \
--body-file .tmp/issue-plan.md
# When there is progress:
gh issue comment <n> --repo openJiuwen-ai/sciencediscovery \
--body-file .tmp/issue-progress.md
Follow create-github-pr. The commands that file the request:
login=$(gh api user --jq .login)
git push -u <fork-remote> HEAD:<branch>
gh pr create --repo openJiuwen-ai/sciencediscovery \
--draft --base main --head "${login}:<branch>" \
--title "type(scope): what changes, in English" \
--body-file .tmp/pr.md
gh pr view <n> --repo openJiuwen-ai/sciencediscovery \
--json number,title,state,isDraft,baseRefName,headRefName,url
Keep the braces in "${login}:<branch>": zsh reads $login:chore/… as
$login with a :c modifier (likewise :h, :t, :r, :e) and sends
GitHub a head that is not your branch. The symptom and both of its causes are
in the create-github-pr skill's Troubleshooting, under No commits between.
Keep it a draft while preparing the change. Drafts already run L1. Once it is ready, mark it ready for review to start UT, ST and mocked E2E, then request review from a code owner listed in .github/CODEOWNERS. An automated review also runs on the pull request.
gh pr ready <n> --repo openJiuwen-ai/sciencediscovery
The title is an English sentence a reader can understand without the diff.
update, fix bug, and a raw file name are not titles. The body states what
changed, why, what was run (with numbers), and the E2E conclusion. A change
with no user-observable path may say the E2E is not applicable and why.
Backend-only is not that reason.
git push origin a task branch.まだレビューはありません。使ってみた感想をお寄せください。
ログインなしで書き込めます。使ってみた感想、質問、動かなかった環境などを気軽にどうぞ。
まだコメントはありません。最初のコメントを書いてみませんか。
Prepare, launch, monitor, and summarize the real RFdiffusion to ProteinMPNN to Protenix antibody pipeline on a local or remote ScienceDiscovery Runner with sandboxed Ascend NPUs.
日本語の概要は準備中です。原文の説明を表示しています。
Independent multi-dimensional assessor for material design candidates. Loads configurable scoring rubrics to evaluate candidates from different expert perspectives. Each dispatch produces structured scores, pros/cons, and verification data without sharing results with other assessors.
日本語の概要は準備中です。原文の説明を表示しています。
Read, diagnose, and change the CI pipeline: GitHub Actions on pull requests, the merge queue, the nightly schedule and the release tag. Use when a job fails, when editing .github/workflows/, when asking which layer runs what, when reproducing a pipeline failure locally, or when a job needs the bubblewrap sandbox. Running the layers before proposing a change, and reading the result a proposal receives, belong to create-github-pr.
日本語の概要は準備中です。原文の説明を表示しています。
Verify cited paper identity and decide whether claims in an Artifact are supported by their linked Evidence using a governed source snapshot.
日本語の概要は準備中です。原文の説明を表示しています。
Use this skill when you need to write and execute Python/R code to process, transform, and analyze data, delivering reproducible computational results with complete code-level methodology documentation. Supports statistical analysis, data transformation, visualization, method justification, and structured result output.
日本語の概要は準備中です。原文の説明を表示しています。
Check whether numeric claims marked with Evidence aliases in an Artifact agree with the linked Evidence content and context.
日本語の概要は準備中です。原文の説明を表示しています。