aivc-curator

name	aivc-curator
description	Use when the user wants to judge whether a paper, dataset, benchmark, blog, or related resource belongs in Awesome-Virtual-Cell, whether an existing README entry should be updated, where an item should be placed, or how it should be written in house style.

AIVC Curator

Help maintainers make consistent inclusion and placement decisions for Awesome-Virtual-Cell.

This skill is repository-specific. It should apply the current README scope and inclusion rules rather than generic biology curation instincts.

When to Use

Use this skill when the user asks:

• whether a paper should be included
• whether a dataset, benchmark, report, or video is in scope
• where an item should live in the README
• whether something is core or only related
• how to write a new README entry in the current house style

Do not use this skill as the main workflow when the user:

• wants general advice on building an awesome project
• wants broad README structure redesign
• needs a live literature search without a curation decision

Workflow

1. Read references/aivc-curation-rules.md.
2. If the task is about a concrete item, check whether the same work already appears in README.md by title, alias, acronym, DOI, or project name.
3. Extract the item type:
   • overview paper
   • research paper
   • dataset
   • benchmark or challenge
   • report or blog
   • video or talk
   • historical or foundational work
   • related resource
4. Extract the maintenance action:
   • new entry
   • update existing entry
   • duplicate or no-op
   • demote, move, or remove
5. Verify the minimum metadata needed for a reliable call:
   • canonical title
   • venue and year
   • primary link
   • optional code, dataset, benchmark, or project link
6. Judge fit against the repository’s scope:
   • core AIVC
   • adjacent but useful
   • weakly related or out of scope
7. Return a structured decision with:
   • include / update-existing / hold / exclude
   • suggested section
   • core vs related
   • duplicate status
   • short rationale
   • exact fields to correct if the recommendation is update-existing
   • missing information to verify if the call is uncertain
   • suggested README entry text if inclusion is recommended
8. If the user asks about a borderline item, explain the boundary instead of forcing a confident decision.
9. Default to Chinese unless the user is writing in English.

Output Contract

Prefer this compact structure:

收录建议：
- 结论：收录 / 更新现有 / 暂缓 / 不收录
- 建议位置：
- 属性：主线 / Related
- 重复检查：

理由：
- ...

若更新现有：
- 需要修改：

还需要确认：
- ...

建议条目写法：
- ...

For English requests, use:

Decision:
- Recommendation: include / update-existing / hold / exclude
- Suggested section:
- Placement: core / related
- Duplicate check:

Rationale:
- ...

If updating existing:
- Fields to fix:

Still needed:
- ...

Suggested README entry:
- ...

Quality Rules

• Follow the repository’s current Scope and Inclusion Rules, not personal preference.
• Prefer primary-source links and higher-confidence resources.
• Prefer updating an existing entry over adding a near-duplicate.
• If something is useful but off the main path, prefer Related Resources over forced placement in the core sections.
• Be conservative with low-confidence sources, broken links, or overly generic biomedical AI resources.
• When recommending inclusion, keep the suggested entry concise and aligned with the README’s existing format.

Reference Map

Load only the files needed for the current curation task:

• references/aivc-curation-rules.md -> executable scope, placement rules, in-scope boundaries, and entry-writing patterns for this repository
• README.md -> duplicate checks, existing wording, year buckets, and section placement
• CONTRIBUTING.md -> contributor-facing submission expectations when the user asks how to submit or format changes