Operate GitCode issues, PRs, wikis, code/MR refs, and cached org templates.

Apache-2.0Auto-check passed

Install Gitcode

skills CLI
$ npx skills add openJiuwen-ai/sciencediscovery --skill gitcode -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install openJiuwen-ai/sciencediscovery gitcode --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/openJiuwen-ai/sciencediscovery.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.agents/skills/gitcode .claude/skills/gitcode && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
gitcode
GitHub stars
151
Token cost
~4.3k tokens
SKILL.md length
1,787 words
Files
4 (incl. scripts)
Skills in repo
22
Repo updated
First seen
Licence
Apache-2.0

At a glance

Operate GitCode issues, PRs, wikis, code/MR refs, and cached org templates.

  • Works in 10 steps: Run gitcode, never bare gc (often a git… → Install/helpers via uv; never commit CLI… → Never print tokens (auth token / auth… → …
  • GitCode issues/PRs
  • SKILL.md covers Rules, Hosts, Install / auth (once) and Targeting, plus 5 more sections
  • Runs Python scripts from its folder; calls uv, git and jq; reaches gitcode.com and raw.gitcode.com; needs GC_TOKEN and GITCODE_TOKEN

What it does

Gitcode is an agent skill from openJiuwen-ai/sciencediscovery. Operate GitCode issues, PRs, wikis, code/MR refs, and cached org templates. Use for gitcode.com, atomgit.com, GitCode issues/PRs, wiki, or /gitcode.

Its SKILL.md is about 4.3k tokens, which your agent loads only when the skill is triggered. The skill folder holds 4 other files, including scripts (for example `scripts/download_image.py`, `scripts/fetch_openjiuwen_templates.py` and `scripts/upload_image.py`).

The repository describes itself as: ScienceDiscovery is an all‑in‑one agentic workbench built specifically for scientific research. The licence is Apache-2.0.

When your agent uses it

  • GitCode issues/PRs

Example prompts

  • “/gitcode”

Requirements

  • Python 3
  • A credential in GITCODE_TOKEN

Workflow steps

10 steps, taken from the first numbered list in SKILL.md.

  1. Run gitcode, never bare gc (often a git commit alias).
  2. Install/helpers via uv; never commit CLI binaries or tokens.
  3. Never print tokens (auth token / auth status --show-token).
  4. Prefer --json; if create/edit JSON is thin (missing html_url or with the wrong number), confirm with view/list.
  5. For a complete Issue inventory or count, never rely on issue list's default --limit 30. Pass an explicit limit larger than the expected…
  6. Destructive ops (close/reopen/merge/delete…) need user intent + --yes.
  7. Scope: only repos the user named (or cwd remote when clearly that clone).
  8. Transports: metadata/API → gitcode CLI; code/MR head → SSH git. HTTPS git easily fails (hangs without a credential helper; web/HTTPS…
  9. Follow Cross-references and issue association for visible links and unverified auto-close behavior.
  10. sciencediscovery merge requests: each operator uses their own GitCode fork of openJiuwen/sciencediscovery (do not hard-code a login)…

What it can do on your machine

Read from SKILL.md and the folder at commit 7a03242. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Ships 3 files in scripts/ (Python), which the agent can run.

    Shell commands in SKILL.md call:

    • uv
    • git
    • jq

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    Hosts in commands or code, which the agent is likely to contact:

    • gitcode.com
    • raw.gitcode.com

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names these keys or tokens, usually read from environment variables:

    • GC_TOKEN
    • GITCODE_TOKEN

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Gitcode loads about 4.3k tokens when it runs. Until then it costs about 39 tokens; SKILL.md has 1,787 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~39
When it runs · the whole SKILL.md, loaded when a task matches
~4.3k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); the scripts in this folder are not scanned.

SKILL.md

The full file from openJiuwen-ai/sciencediscovery at commit 7a03242, republished under its Apache-2.0 licence (© openJiuwen-ai). 1,787 words, ~4,295 tokens.

Download SKILL.mdSave it as .claude/skills/gitcode/SKILL.md (or your agent's skills folder). This skill also uses 3 other files; get the full folder from GitHub.
name
gitcode
description
Operate GitCode issues, PRs, wikis, code/MR refs, and cached org templates. Use for gitcode.com, atomgit.com, GitCode issues/PRs, wiki, or /gitcode.

GitCode (Issue / PR)

Project-local skill for sciencediscovery on gitcode.com (not under skills/).

New issues and pull requests for this repository are not filed here. Use gh against openJiuwen-ai/sciencediscovery, and write the title and body in English. This skill is for an existing GitCode mirror item that still has to be read or commented on.

Use the command table below, gitcode schema, or gitcode <cmd> --help to resolve flags. Only if a flag is still unclear, run gitcode api repos/gitcode-cli/cli/contents/docs, select the download_url for COMMANDS.md, AUTH.md, or AI-GUIDE.md, and read that raw.gitcode.com content. Do not scrape GitCode blob-page HTML.

Rules

  1. Run gitcode, never bare gc (often a git commit alias).
  2. Install/helpers via uv; never commit CLI binaries or tokens.
  3. Never print tokens (auth token / auth status --show-token).
  4. Prefer --json; if create/edit JSON is thin (missing html_url or with the wrong number), confirm with view/list.
  5. For a complete Issue inventory or count, never rely on issue list's default --limit 30. Pass an explicit limit larger than the expected result set, check the returned JSON array length, and continue with --page 2, --page 3, … whenever a page length equals the requested limit. Stop only after a shorter page, then deduplicate by Issue number before counting.
  6. Destructive ops (close/reopen/merge/delete…) need user intent + --yes.
  7. Scope: only repos the user named (or cwd remote when clearly that clone).
  8. Transports: metadata/API → gitcode CLI; code/MR head → SSH git. HTTPS git easily fails (hangs without a credential helper; web/HTTPS fetches commonly return 403), so always use SSH and do not scrape gitcode.com HTML.
  9. Follow Cross-references and issue association for visible links and unverified auto-close behavior.
  10. sciencediscovery merge requests: each operator uses their own GitCode fork of openJiuwen/sciencediscovery (do not hard-code a login). Resolve <gitcode-login> from gitcode auth status --json. Push the source branch to that fork (common remote name gitcode-fork) and create with --head <gitcode-login>:<branch>. Do not git push origin <task-branch> onto openJiuwen/sciencediscovery. If the fork is missing: gitcode repo fork openJiuwen/sciencediscovery --json.

Hosts

gitcode.com and atomgit.com are hostnames for the same website/platform: repositories and their issue/PR numbers are identical. Do not treat AtomGit as a separate product. For authored or reported repo, issue, and PR URLs, prefer https://gitcode.com/...; rewrite AtomGit hosts and CLI html_url paths such as /merge_requests/N to these canonical forms:

  • Issue: https://gitcode.com/openJiuwen/sciencediscovery/issues/9
  • PR list: https://gitcode.com/openJiuwen/sciencediscovery/pulls
  • PR detail: https://gitcode.com/owner/repo/pull/N
  • SSH: git@gitcode.com:owner/repo.git and git@atomgit.com:owner/repo.git are equivalent; prefer writing/configuring the gitcode.com form, even when an existing remote uses AtomGit.

Install / auth (once)

bash
uv tool install gitcode-cli
export PATH="$HOME/.local/bin:$PATH"
gitcode version
gitcode auth status --json   # need logged_in + token_valid

Token priority: GC_TOKEN > GITCODE_TOKEN > ~/.config/gc/auth.json. No browser on box → user creates classic token at https://gitcode.com/setting/token-classic and runs gitcode auth login on a private TTY (do not ask them to paste token into chat). Fallback install: uv venv .local/gitcode-venv && uv pip install --python .local/gitcode-venv gitcode-cli. Binary resolve: command -v gitcode → $HOME/.local/bin/gitcode → .local/gitcode-venv/bin/gitcode. PyPI stale: uv tool install --from 'https://gitcode.com/gitcode-cli/cli/releases/download/<ver>/gitcode_cli-<ver>-py3-none-any.whl' gitcode-cli.

Targeting

Always pass -R owner/repo when not on cwd origin. Forms: owner/repo, HTTPS URL, SSH URL (CLI API only — not git transport).

NeedUse
Title/body/state/comments/create/edit/reviewgitcode …
Patch summarygitcode pr diff N (text preferred; binaries have no preview)
Full tree / deep reviewSSH remote git@gitcode.com:owner/repo.git; fetch refs/merge-requests/N/head
Scratch clone.tmp/<repo-name>/
Long bodies.tmp/*.md + --body-file

Issue / PR commands

bash
# issues
gitcode issue create -R owner/repo --title "…" --body-file body.md --json
# Complete inventory: 30 is the CLI default and is commonly truncated.
gitcode issue list -R owner/repo --state open --limit 100 --page 1 --json
gitcode issue view N -R owner/repo --json
gitcode issue view N -R owner/repo --comments --json
gitcode issue edit N -R owner/repo --body-file body.md --json
gitcode issue comment N -R owner/repo --body "…" --json
gitcode issue comments N -R owner/repo --json
gitcode issue comment edit <comment_id> -R owner/repo --body "…"   # may not support --json
gitcode issue close N -R owner/repo --yes --json
gitcode issue reopen N -R owner/repo --yes --json

# PRs — sciencediscovery: source branch on the operator's own fork, never on origin
gitcode pr create -R openJiuwen/sciencediscovery --head <gitcode-login>:branch --base main --title "…" --body-file pr.md --json
gitcode pr create -R upstream/repo --head myfork:branch --title "…" --body-file pr.md --json
gitcode pr create -R owner/repo --head branch --fill --json
gitcode pr create -R owner/repo --head branch --title "WIP" --draft --json
gitcode pr list -R owner/repo --state open --json
gitcode pr view N -R owner/repo --json
gitcode pr view N -R owner/repo --comments --json
gitcode pr diff N -R owner/repo
gitcode pr edit N -R owner/repo --body-file pr.md --json
gitcode pr comment N -R owner/repo --body "…" --json
gitcode pr comment N -R owner/repo --body "…" --path path/to/file.py --position 12 --json
gitcode pr comments N -R owner/repo --json
gitcode pr reply N -R owner/repo --discussion <id> --body "…"
gitcode pr review N -R owner/repo --comment "…" --json   # or --comment-file
gitcode pr merge N -R owner/repo --yes --json            # optional --method squash|rebase
gitcode pr close|reopen|ready N -R owner/repo --yes --json
gitcode pr checkout N -R owner/repo
gitcode repo fork owner/repo --json
gitcode repo view owner/repo --json
gitcode schema
gitcode schema "issue create"
PR view JSON shapes

gitcode pr view changes its top-level JSON shape when comments are requested:

  • gitcode pr view N -R owner/repo --json returns the PR object directly.
  • gitcode pr view N -R owner/repo --comments --json returns { "pull_request": <PR object>, "comments": [...] }.

Do not read .number, .head, or .labels at the top level of the comments form; those fields are below .pull_request. Optional arrays such as labels may be null, so normalize them with // [] before iterating.

bash
# PR metadata only
gitcode pr view N -R owner/repo --json |
  jq '{number, title, state, head_sha: .head.sha,
       labels: [((.labels // [])[]) | .name]}'

# PR metadata and comments
gitcode pr view N -R owner/repo --comments --json |
  jq '{pr: (.pull_request | {number, title, state, head_sha: .head.sha}),
       comments: [(.comments // [])[] |
         {author: (.user.login // .author.login // .author_name),
          body: (.body // .note // ""), created_at}]}'

Use gitcode pr comments N -R owner/repo --json when only the comment list is needed.

For complete Issue inventories, treat length == --limit as “possibly truncated,” not as a final count. Fetch subsequent pages with the same filters until one returns fewer rows than the limit; combine the pages and deduplicate by .number. If the first page returns fewer than the explicit limit, it is complete for those filters at that retrieval time.

  • Line comments: --position = line on the new file (right side of diff). Inline comments appear as comment_type: diff_comment.
  • --approve: only with explicit user intent + approval permission; own-PR / missing role → 403 — leave a comment review instead.
  • --request (request changes) unsupported; put change requests in --comment.
  • Merge only with explicit user OK for that PR.
  • pr create 409 / "same source branch already has an open MR !N": !N may be the PR this very call just created (observed on this repo), an earlier PR opened from the same branch, or one you should not touch. Never retry blindly and never rename the branch to dodge it. Read it back — gitcode pr view N -R owner/repo --json — compare base, head branch, head SHA, title, body, author, created time, and existing review activity against this attempt, then act on what !N actually is. Typical outcomes: this attempt landed ⇒ report !N; your own re-delivery on the same branch, where the MR already tracks the pushed head but title/body are stale ⇒ bring it up to date with gitcode pr edit N --body-file … instead of creating a second PR; head SHA still behind your push ⇒ confirm the push reached the fork before touching the PR; wrong base, another author, or review already under way ⇒ stop and report so the user can choose (edit, close, or a fresh branch). Judge from the read-back, not from the error text.
Read a PR (review prep)
  1. gitcode pr view N -R owner/repo --json — title, body, SHAs, state
  2. gitcode pr diff N -R owner/repo — patch overview; prefer text output, and expect no binary preview
  3. gitcode api repos/owner/repo/pulls/N/files — structured file list / raw URLs
  4. Full tree: SSH git fetch git@gitcode.com:owner/repo.git refs/merge-requests/N/head (GitCode MR ref; not HTTPS). Prefer this over scraping the web UI.
Cross-references and issue association
  • Ordinary issue/comment reference: [#32](https://gitcode.com/openJiuwen/sciencediscovery/issues/32).
  • Ordinary PR reference: [#10](https://gitcode.com/openJiuwen/sciencediscovery/pull/10). Write /pull/N in authored links even if CLI html_url says /merge_requests/N.
  • To associate an issue in a PR body, retain the trigger candidate and add a readable link: Fixes #32 ([#32](https://gitcode.com/openJiuwen/sciencediscovery/issues/32)).
  • Automatic closing after merge is not verified. Never promise that Fixes #N or its Markdown link will close the issue; commit messages alone do not auto-close.
Show full SKILL.md (749 more words)Show less

openJiuwen org templates

Template source order when filing an Issue/PR:

  1. This repo's own .gitcode/ Issue/PR templates, if present (openJiuwen/sciencediscovery currently has none).
  2. Otherwise the openJiuwen org templates below — this is the default source, not an opt-in path.

Language: a GitHub issue or pull request is English and is created with gh — see github. This section applies only when you are still editing an item on the GitCode mirror. There, default to the Chinese templates (ISSUE_TEMPLATE.zh/ + PULL_REQUEST_TEMPLATE.md / PULL_REQUEST_TEMPLATE.zh-CN.md). Use the English set (ISSUE_TEMPLATE.en/, PULL_REQUEST_TEMPLATE.en.md) only when the person who asked explicitly wants English on that mirror item.

Upstream (do not vendor into git): openJiuwen/.gitcode @ master, tree .gitcode/ (web: https://gitcode.com/openJiuwen/.gitcode). Local cache (gitignored): {baseDir}/cache/openjiuwen-org-templates/.

bash
FETCH="{baseDir}/scripts/fetch_openjiuwen_templates.py"
uv run --no-project "$FETCH"          # ensure cache exists
uv run --no-project "$FETCH" --force  # refresh after upstream changes

Layout is split per locale — there is no single ISSUE_TEMPLATE/ directory:

.gitcode/ISSUE_TEMPLATE.zh/*.yml        # Chinese issue forms (default)
.gitcode/ISSUE_TEMPLATE.en/*.yml        # English issue forms (+ config.yml)
.gitcode/PULL_REQUEST_TEMPLATE.md
.gitcode/PULL_REQUEST_TEMPLATE.zh-CN.md
.gitcode/PULL_REQUEST_TEMPLATE.en.md

Do not rely on any template inventory in this skill — lists go stale. At filing time:

  1. Ensure cache is present (run fetch if needed).
  2. List and read the actual files under …/cache/openjiuwen-org-templates/.gitcode/ (ISSUE_TEMPLATE.zh/, ISSUE_TEMPLATE.en/, PULL_REQUEST_TEMPLATE*). FETCH_META.json records what the last fetch wrote (issue_template_dirs, pr_templates), but read the files themselves before filling one in.
  3. Pick the matching template for the intent from the language directory chosen above; fill required fields from that file; create with --body-file.
  4. If the issue template is an issue form (YAML), convert it to Markdown in body order: write each attributes.label as a ### heading and put the answer below it; every item with validations.required: true is mandatory.
  5. config.yml is issue-form chooser config, not a template — never file it as an issue body.

Images in bodies/comments

bash
UPLOAD="{baseDir}/scripts/upload_image.py"
uv run --no-project "$UPLOAD" -R owner/repo .tmp/image.png --json
  1. List the source directory first to confirm the actual filename; do not retype a path from memory. Then copy archived/non-ASCII paths to a simple ASCII name under .tmp/ before upload.

  2. Prefer --json; continue only when the command exits 0 and stdout is a single JSON object with non-empty markdown and url fields (success prints no progress). The URL must use https://raw.gitcode.com/user-images/assets/…, never bare /uploads/…. Errors/warnings go to stderr only.

  3. On non-zero exit or stderr containing error:, stop. Never paste stderr/error text into an issue/PR body or replace a failed image placeholder with an empty string.

  4. Build comments and long bodies in .tmp/*.md without shell expansion (use an editor/patch or a single-quoted heredoc), then pass the file with --body-file; do not use $(cat <<EOF) around Markdown.

bash
gitcode pr comment N -R owner/repo --body-file .tmp/comment.md --json
  1. After posting, read back with gitcode issue view N -R owner/repo --comments --json or gitcode pr comments N -R owner/repo --json. Confirm the comment contains the expected raw.gitcode.com embed and contains no error: file not found.
Fetching images (raw.gitcode.com)

raw.gitcode.com user-images URLs require auth — anonymous GET returns 403 no access right (verified platform behavior, not a broken link). Use the skill's download script, which reuses the same token resolution as upload_image.py and sends Authorization: Bearer <token>:

bash
DOWNLOAD="{baseDir}/scripts/download_image.py"
# Bare URL or Markdown embed ![alt](url 'title'); saves to .tmp/<filename> by default
uv run --no-project "$DOWNLOAD" "https://raw.gitcode.com/user-images/assets/<repo_id>/<uuid>/<file>" -o .tmp/issue44_img1.png --json
uv run --no-project "$DOWNLOAD" "![alt](https://raw.gitcode.com/user-images/assets/.../x.png 'x')" -o .tmp/x.png
  1. The script accepts a bare https://raw.gitcode.com/... URL or a full Markdown embed ![alt](url 'title') (e.g. copied straight from an issue body) and extracts the URL.
  2. Prefer --json; continue only when the command exits 0 and stdout is a single JSON object with "success": true, status: 200, and a content_type starting with image/ (success prints no progress). The saved file path is in path.
  3. Default output is .tmp/<url-filename> if .tmp/ exists, else ./<url-filename>; pass -o to override. Non-ASCII filenames (e.g. 企业微信截图_*.png) are kept as-is — copy to an ASCII name before review if needed.
  4. On non-zero exit or stderr containing error:, stop. Exit 4 means auth/403 (token missing, invalid, or expired — run gitcode auth status --json); exit 1 is a transport/server error. Never treat a 403 as a broken image link.
  5. To inspect images referenced in an issue/PR body, gitcode issue view N --json / gitcode pr view N --json and extract raw.gitcode.com URLs from the body field before downloading.

Wiki

No gitcode wiki subcommand. Wiki is owner/repo.wiki (SSH git@gitcode.com:owner/repo.wiki.git), default branch main, entry Home.md. Prefer SSH.

Checklist

  1. gitcode version + auth status --json
  2. Lock -R
  3. Read before write (view/list/diff; code via SSH MR ref) — no HTML scrape
  4. Follow the cross-reference rules above; do not promise unverified auto-close behavior
  5. Confirm number/URL after create/edit; report /pull/N or /issues/N
  6. Failures: stderr + exit class (0 ok, 1 generic, 2 usage, 3 not found, 4 auth, 5 conflict) — conflict (e.g. pr create 409) means read back the cited !N and decide from its actual state, not from the message; never invent success

Policy: skill sources under .agents/skills/gitcode/ are tracked; cache/ is not. No binaries, no vendored templates, no secrets in skill files.

© openJiuwen-ai, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 3 other files (scripts) in .agents/skills/gitcode of openJiuwen-ai/sciencediscovery.

  • SKILL.md
  • scripts/download_image.py
  • scripts/fetch_openjiuwen_templates.py
  • scripts/upload_image.py

Open the folder on GitHubat commit 7a03242

Compare with similar skills

Gitcode next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Gitcode compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Gitcode this skillopenJiuwen-ai/sciencediscovery151—~4.3kAutomated safety check: PassApache-2.0
Wiki Querypaperclipai/paperclip99k—~682Automated safety check: PassMIT
Wiki Maintaineropenclaw/openclaw392k1 repos~462Automated safety check: PassMIT
Wiki Lintpaperclipai/paperclip99k—~878Automated safety check: PassMIT
Wiki Ingestpaperclipai/paperclip99k—~933Automated safety check: PassMIT
LLM WikiYeachan-Heo/oh-my-claudecode40k—~721Automated safety check: PassMIT

Similar skills

  • Wiki Query

    paperclipai/paperclip

    A skill your agent uses when an LLM Wiki operation issue asks a question.

    99k GitHub stars~682 tokensUpdated today
    Knowledge ManagementAuto-check passed
  • Wiki Maintainer

    openclaw/openclaw

    Maintain the OpenClaw memory wiki vault with deterministic pages, managed blocks, and source-backed updates.

    392k GitHub starsUsed in 1 repo~462 tokens
    Knowledge ManagementAuto-check passed
  • Wiki Lint

    paperclipai/paperclip

    A skill your agent uses when an LLM Wiki operation issue is a lint or health check.

    99k GitHub stars~878 tokensUpdated today
    Knowledge ManagementAuto-check passed
  • Wiki Ingest

    paperclipai/paperclip

    A skill your agent uses when an operation issue asks to ingest a captured raw/ source into the LLM Wiki, or the user says "ingest <slug".

    99k GitHub stars~933 tokensUpdated today
    Knowledge ManagementAuto-check passed
  • LLM Wiki

    Yeachan-Heo/oh-my-claudecode

    Keeps a persistent markdown wiki of project and session knowledge that the agent can ingest into, query, lint and read across sessions.

    40k GitHub stars~721 tokensUpdated today
    Knowledge ManagementAuto-check passed
  • Shepherd PRs

    flutter/flutter

    Automate shepherding, checking status, updating branches, and landing open PRs or approved third-party contributor PRs in the flutter/flutter repository using the gh CLI.

    179k GitHub stars~1.7k tokensUpdated today
    MobileAuto-check passed

More from openJiuwen-ai/sciencediscovery

All 22 skills in this repo
  • Code Engineer

    openJiuwen-ai/sciencediscovery

    A skill your agent uses 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…

    151 GitHub stars~2.8k tokensUpdated yesterday
    Auto-check passed
  • Structure Pocket Inspection

    openJiuwen-ai/sciencediscovery

    Inspect a local PDB structure, summarize chains and residue composition, and identify protein atoms near a user-specified ligand or pocket center.

    151 GitHub stars~624 tokensUpdated yesterday
    Auto-check passed
  • Antibody Design

    openJiuwen-ai/sciencediscovery

    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.

    151 GitHub stars~2.9k tokensUpdated yesterday
    Auto-check passed
  • Science Research Team

    openJiuwen-ai/sciencediscovery

    A skill your agent uses to orchestrate a multi-domain research team for literature/evidence research and data analysis.

    151 GitHub stars~1.3k tokensUpdated yesterday
    Auto-check passed
  • Literature Searcher

    openJiuwen-ai/sciencediscovery

    A skill your agent uses when a research workflow needs verified academic source retrieval through literature-search MCP interfaces available in the current session before evidence extraction.

    151 GitHub stars~5.8k tokensUpdated yesterday
    Auto-check passed
  • Create GitHub PR

    openJiuwen-ai/sciencediscovery

    Open or update a pull request on GitHub's openJiuwen-ai/sciencediscovery: run the UT/ST/E2E layers locally, push the branch to the operator's own GitHub fork, write a body that says what was…

    151 GitHub stars~3k tokensUpdated yesterday
    Auto-check passed

Questions about Gitcode

What does Gitcode do?

Operate GitCode issues, PRs, wikis, code/MR refs, and cached org templates. Gitcode is an agent skill from openJiuwen-ai/sciencediscovery. Operate GitCode issues, PRs, wikis, code/MR refs, and cached org templates.

When should I use Gitcode?

Gitcode fits situations like: gitCode issues/PRs.

How do I install Gitcode in Claude Code?

Run `npx skills add openJiuwen-ai/sciencediscovery --skill gitcode -a claude-code`. Or copy the skill folder (.agents/skills/gitcode in openJiuwen-ai/sciencediscovery) into .claude/skills/gitcode in your project. Claude Code loads it when a task matches its description.

How do I install Gitcode in Codex?

Run `npx skills add openJiuwen-ai/sciencediscovery --skill gitcode -a codex`. Or copy the skill folder (.agents/skills/gitcode in openJiuwen-ai/sciencediscovery) into .agents/skills/gitcode in your project. Codex loads it when a task matches its description.

Can I use Gitcode in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add openJiuwen-ai/sciencediscovery --skill gitcode -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/gitcode, .gemini/skills/gitcode, .github/skills/gitcode and .opencode/skills/gitcode in your project.

What does Gitcode need to run?

Going by SKILL.md and its folder, Gitcode needs Python for the scripts in its folder, the command-line tools its instructions call (uv, git and jq) and credentials named GC_TOKEN and GITCODE_TOKEN. Our summary lists: Python 3; A credential in GITCODE_TOKEN.

Does Gitcode access the network?

SKILL.md names 2 domains. In commands or code: gitcode.com and raw.gitcode.com; the agent is likely to contact these when it follows the instructions. This is read from the text; nothing was executed.

Is Gitcode safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.

What licence does Gitcode use?

Gitcode is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Gitcode use?

About 4.3k tokens (SKILL.md is roughly 17k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Gitcode?

Skills that share tags, products or a category with Gitcode: Wiki Query (paperclipai/paperclip, 99k stars), Wiki Maintainer (openclaw/openclaw, 392k stars), Wiki Lint (paperclipai/paperclip, 99k stars) and Wiki Ingest (paperclipai/paperclip, 99k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Gitcode?

openJiuwen-ai (a GitHub organization) maintains it in openJiuwen-ai/sciencediscovery, which has 151 GitHub stars. The repository holds 22 skills in this directory. The repository was last updated on October 7, 2026.

Source: openJiuwen-ai/sciencediscovery on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.