Agent skill

Research Literature Radar

by huangwb8 in huangwb8/ChineseResearchLaTeX

发现、筛选并长期归档重要研究论文;当用户要求按主题寻找经典、rising star、社区精选、热点或顶会/顶刊论文时使用。采用分层发现策略,调用 research-literature-search 完成其中的关键词/数据库检索,再由本 skill 负责多渠道汇总、idea-level 价值判断、论文类型分类、跨轮次跟踪和论文库归档。

MITAuto-check passedResearch & Science

Install Research Literature Radar

skills CLI
$ npx skills add huangwb8/ChineseResearchLaTeX --skill research-literature-radar -a claude-code

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

GitHub CLI
$ gh skill install huangwb8/ChineseResearchLaTeX research-literature-radar --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/huangwb8/ChineseResearchLaTeX.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/research-literature-radar .claude/skills/research-literature-radar && 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
research-literature-radar
GitHub stars
2.9k
Token cost
~1.5k tokens
SKILL.md length
289 words
Files
11 (incl. scripts)
Skills in repo
27
Repo updated
First seen
Licence
MIT

At a glance

发现、筛选并长期归档重要研究论文;当用户要求按主题寻找经典、rising star、社区精选、热点或顶会/顶刊论文时使用。采用分层发现策略,调用 research-literature-search 完成其中的关键词/数据库检索,再由本 skill 负责多渠道汇总、idea-level 价值判断、论文类型分类、跨轮次跟踪和论文库归档。

  • Tasks that involve Literature review
  • SKILL.md covers 目标, 流程 and 约束
  • Runs Python scripts from its folder; calls gh

What it does

Research Literature Radar is an agent skill from huangwb8/ChineseResearchLaTeX. 发现、筛选并长期归档重要研究论文;当用户要求按主题寻找经典、rising star、社区精选、热点或顶会/顶刊论文时使用。采用分层发现策略,调用 research-literature-search 完成其中的关键词/数据库检索,再由本 skill 负责多渠道汇总、idea-level 价值判断、论文类型分类、跨轮次跟踪和论文库归档。

Its SKILL.md is about 1.5k tokens, which your agent loads only when the skill is triggered. The skill folder holds 12 other files, including scripts (for example `CHANGELOG.md`, `README.md` and `agents/openai.yaml`).

It sits in Research & Science, covering Literature review. The licence is MIT.

When your agent uses it

  • Tasks that involve Literature review

Example prompts

  • “/research-literature-radar”

Requirements

  • Python 3

What it can do on your machine

Read from SKILL.md and the folder at commit b8b4142. 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 6 files in scripts/ (Python), which the agent can run.

    Shell commands in SKILL.md call:

    • gh

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

  • Network

    No URLs in SKILL.md. Its commands use gh, which can reach the network depending on how they are called.

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

  • Credentials

    Names no API keys, tokens, secrets or passwords.

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

Context cost

Research Literature Radar loads about 1.5k tokens when it runs. Until then it costs about 49 tokens; SKILL.md has 289 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~49
When it runs · the whole SKILL.md, loaded when a task matches
~1.5k

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 huangwb8/ChineseResearchLaTeX at commit b8b4142, republished under its MIT licence (© huangwb8). 289 words, ~1,539 tokens.

Download SKILL.mdSave it as .claude/skills/research-literature-radar/SKILL.md (or your agent's skills folder). This skill also uses 10 other files; get the full folder from GitHub.
name
research-literature-radar
description
发现、筛选并长期归档重要研究论文;当用户要求按主题寻找经典、rising star、社区精选、热点或顶会/顶刊论文时使用。采用分层发现策略,调用 research-literature-search 完成其中的关键词/数据库检索,再由本 skill 负责多渠道汇总、idea-level 价值判断、论文类型分类、跨轮次跟踪和论文库归档。
metadata.author
Bensz Conan

Research Literature Radar

目标

把来自不同发现渠道的候选论文转成可持续维护的研究雷达。research-literature-search 只是本 skill 的一个检索子步骤,负责基于显式关键词的数据库召回、字段规范化、canonical 去重和 provenance;本 skill 仍负责分层发现、跨渠道汇总、价值判断、分类、跟踪与归档。

本 skill 负责:

  • 将用户目标转成筛选标准和论文类型配额;
  • 对全部 canonical 候选分别判断相关性、证据深度、发表状态和身份可信度;
  • 分类、排序、记录不确定性,并保留核心、辅助、待跟踪和范围外角色;
  • 将入选论文映射到稳定 ID,写入论文库并维护跨轮次跟踪。
触发与输入概览

用户提出“找值得读的论文”“建立某主题论文雷达”“按经典/热点/顶会等类型推荐论文”或类似发现与学习需求时触发。

必填:用户选题和目标数量。可选:领域/子主题、五类论文配额(classic、rising-star、community、hot、top-venue)、时间窗、作者/venue 白名单、排除项、是否下载公开 PDF、笔记深度。

运行前读取 .bensz-api/research-literature-radar/catalog.jsonl(不存在则创建),并保留用户原始请求和运行日期。

流程

输入

按用户请求和配置文件提供必要输入;缺失信息应明确列出并停止依赖该输入的步骤。

执行步骤
  • 仅将本 Skill 的设计缺陷(流程漏判、输入契约不完整或环境假设错误)视为可上报 bug;用户数据错误、第三方服务抖动、用户主动改源码和模型偶发波动不属于此范围。
  • 发现设计缺陷时先脱敏记录到 ~/.bensz-skills/bugs/,当前任务继续;只有用户明确要求公开上报时,才使用本机 gh api 直传,不 clone 仓库。
  • 不收集用户名、主机名、工作目录、密钥、令牌、Cookie 或其它无关隐私;不得直接修改用户本地已安装 Skill 的源代码来“顺手修 bug”。

版本唯一来源为同目录 config.yaml:skill_info.version;本文件只描述稳定工作契约,不重复易变配置。

不要把论文发现简化成单一关键词检索。根据用户选题和目标数量,组合以下五类渠道;每类都可以使用联网检索、研究者脉络或社区信号,但必须记录其来源和用途:

  • 经典论文:定位领域奠基者、明星研究者、综述、奖项和公认里程碑,再查找其代表性工作。
  • rising-star 论文:寻找近 3–5 年持续突破的作者、实验室或研究线,关注尚未广为人知但影响快速上升的研究者。
  • 社区精选:检查 Hugging Face Daily/Weekly/Monthly Papers、Import AI、研究者通讯、公众号/X 账号和可信整理页。
  • 偶然的热点论文:关注近期发表后因讨论度、代码传播、新闻或社交平台而受到关注的论文,即使作者并不知名。
  • 顶会/顶刊近期论文:仅在用户选题足够窄时重点使用,结合明确的 venue 和时间窗,避免宽主题造成无穷候选。

research-literature-search 负责其中“基于关键词检索学术数据库”的小步骤;经典作者脉络、社区精选和热点信号不能假设会自动出现在该检索 bundle 中。

首选:调用 search 后端

当分层策略需要关键词/数据库召回时,调用 research-literature-search 的 run,为其提供合法的 topic 和 5–25 条显式查询;随后调用其 validate。只接受 manifest.json 的 status 为 success 或 partial_success 的 bundle。没有需要数据库检索的渠道时,不得为了形式强行调用;但必须说明使用了哪些其它发现渠道。

将以下文件作为只读输入保存到本轮 radar run:

text
manifest.json
candidates_deduped.jsonl
provenance.jsonl
dedupe_map.json

候选必须符合 rls.paper.v1。保留 manifest、contract version、artifact hash、canonical 候选数量和 search source path,便于追溯。若实际调用了 search 但 bundle 校验失败、没有合法查询或 search skill 不可发现,应停止该检索子步骤并报告原因;不能静默改用另一套内嵌 provider。其它分层渠道仍可独立记录为补充发现,但不得伪装成 search 结果。

补充发现信号

社区精选、奖项、研究者整理、代码传播或新闻讨论等非标准数据库信号可以作为补充证据,但必须:

  • 记录 URL、访问日期、来源用途和对应论文身份;
  • 映射到已有 canonical 候选,或作为明确标记的待确认候选;
  • 不伪装成 search provider 结果,不覆盖 search 的 canonical 顺序和去重结论。

当补充信号产生新论文时,使用本 skill 的稳定身份键与 catalog 比对;相似但无法确认的记录标记 possible_duplicate,不得静默合并。

对全部 canonical 候选先建立 landscape,再对可能进入核心集者评分。每条记录包含 record_id、canonical_rank、role、cluster、relevance、evidence_depth、publication_status、identity_confidence、use_cases、reason、uncertainties 和 source_refs。标题可用于发现、聚类和 watchlist;摘要可支持相关性与作者自报判断;方法、结果和等价性强结论必须匹配全文证据。预印本状态与来源数量不能替代相关性判断。

由 AI 批量聚类并填写简短判断草稿,再运行 scripts/build_landscape.py --bundle <search-bundle> --draft <draft.jsonl> --output <literature-landscape.jsonl> --summary <landscape-summary.json> 做 manifest/hash、全量覆盖、ID 唯一、枚举、canonical 顺序与计数对账。确定性脚本不得硬编码领域关键词或替 AI 判定科学价值。

对核心候选按 0–5 分记录:conceptual_novelty、simplicity、surprise、generality、unification、new_primitive、follow_up_potential、practical_impact。证据质量单列为 confidence,不混入“有趣程度”。

为每篇核心候选记录总分、两句理由、关键证据、思想标签和不确定性;其余候选仍保留在 landscape,不能缩成落选总数。思想标签可使用:problem-reformulation、unexpected-simplicity、hidden-equivalence、assumption-revisit、new-measurement、failure-revealing、new-primitive、cross-domain-transfer。

用配额检查研究线覆盖,但不为填满配额选择低质量论文;再按总分、思想标签/年份/作者多样性形成精而少的核心集。辅助或待跟踪条目成为关键近邻时保留 Search record ID,并可新增 R 锚点和解读历史,不重新去重。

scripts/catalog.py 提供不依赖网络的 ID 生成、标题标准化、索引加载和重复判定;scripts/validate_layout.py 在交付前检查运行级文件是否错误写入 docs/papers/ 根目录。脚本不会下载或删除文件。

输出

每轮先加载 catalog,再消费已验证的 search bundle。重复项只更新来源、版本、评分或跟进记录;只有真正新论文才创建目录。将 literature-landscape.jsonl、landscape-summary.json、discovery.md、selection.md、dedup-report.md 和运行摘要写入同一 runs/<run-id>/,并明确 canonical 数 = core + supporting + watchlist + out_of_scope。

交付前确认:每个实际使用的 search manifest 可消费、canonical hash 与数量可追溯;landscape 恰好覆盖每个 canonical record ID;五类渠道的适用性和覆盖情况有说明。一个可靠来源通常足以确认身份;只有元数据冲突、版本合并、疑似重复或决定性近邻才要求额外核对。ID、metadata、catalog、raw manifest 与内部路径一致;根目录 index.md 已包含每篇新增论文;docs/papers/ 根目录没有运行级文件。

输出管理

目录 ID 使用 <first-author-full-name>-<year>-<keyword>:首位作者全名、四位年份和可识别工作/方法关键词均为 ASCII 小写并以单个连字符分隔。关键词优先使用公认简称;无稳定简称时使用标题短 slug。arXiv、DOI、OpenReview 等外部编号不能作为目录主名,必须写入 identifiers。

身份比对顺序为 DOI → arXiv/OpenReview → 其它稳定 ID → 标准化标题+首位作者+年份。预印本与正式版合并并保留版本链接;冲突时在关键词末尾追加稳定短哈希。已有笔记不覆盖,只补充缺失字段并在 history 写明变更原因。

text
docs/papers/
└── <friendly-id>/
    ├── raw/                       # 公开 PDF、HTML、metadata、manifest
    └── <friendly-id>.md           # 学习笔记

.bensz-api/research-literature-radar/
├── catalog.jsonl                  # 跨轮次论文索引
└── runs/<run-id>/                 # run.yaml、search 交接包、筛选与去重报告

docs/papers/ 只存论文实体及其原始材料/学习笔记;运行级配置、manifest、筛选报告、日志和去重报告统一写入 .bensz-api/research-literature-radar/。原始文件用外部标识命名,可记录 SHA-256,禁止静默覆盖。

catalog.jsonl 每行至少记录 id,title,authors,year,venue,status,paper_types,tags,identifiers,sources,files,scores,confidence,first_seen_run,last_seen_run,history,且 id 必须与论文目录名和笔记文件名一致。

校验

完成后执行 Skill 已有的静态检查、脚本验证或人工复核,并记录通过标准。

失败与恢复

保留错误证据和已完成产物;仅在输入、环境或外部依赖恢复后从最近的失败步骤重试。

约束

遵守以下公共约束,并执行本 Skill 的专属边界。

公共硬约束
  • 任务需要落盘时,使用唯一的 ./.bensz-api/task-{yyyymmdd-hhmm}-{简短描述}/ 根目录;共享材料放入 shared/,Skill 专属材料放入该 Skill 的 input/、output/、log/。
  • 正式交付物、源代码和正式计划按项目约定保存,不写入任务工作区;未经授权不覆盖、删除、迁移或远程写入。
  • 项目维护变更检查 BAC 可用性并记录需求、AI 产出、工具结果、文件改动和验证摘要;BAC 只做过程审计,不替代署名、责任或合规判断。
  • 不记录 API Key、访问令牌、密码、Cookie、环境/凭据文件、私有 Prompt、身份信息、本地用户名、主机名或不必要的大体积原始数据。
  • 文件路径必须规范化并限制在授权项目范围内;外部 URL、子进程和网络访问遵循最小权限,防止路径遍历、SSRF 和命令注入。
  • Skill 版本唯一记录在自身 config.yaml:skill_info.version;公开 API、协议、目录或配置变更同步文档与 CHANGELOG.md。
  • 仅将 Skill 或 Bensz 基础设施本身的设计缺陷交给 bensz-collect-bugs;先脱敏写入 ~/.bensz-skills/bugs/,当前任务不中断,只有用户明确要求才公开上报,禁止直接修改用户已安装的 Skill 源码。
<!-- End of canonical common constraints. -->
Skill 专属约束

不得超出本 Skill description 和上方流程所声明的范围;不将未验证的信息伪装成确定结论。

© huangwb8, MIT. 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 10 other files (scripts) in skills/research-literature-radar of huangwb8/ChineseResearchLaTeX.

  • SKILL.md
  • CHANGELOG.md
  • README.md
  • agents/openai.yaml
  • config.yaml
  • scripts/build_landscape.py
  • scripts/catalog.py
  • scripts/finalize_corpus.py
  • scripts/generate_initial_corpus.py
  • scripts/reconcile_integrity.py
  • scripts/validate_layout.py

Open the folder on GitHubat commit b8b4142

Compare with similar skills

Research Literature Radar 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.

Research Literature Radar compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Research Literature Radar this skillhuangwb8/ChineseResearchLaTeX2.9k—~1.5kAutomated safety check: PassMIT
Nature Paper CardYuan1z0825/nature-skills47k2 repos~2.1kAutomated safety check: PassApache-2.0
Systematic Review ScreenerImbad0202/academic-research-skills51k—~8.4kAutomated safety check: PassCustom licence
Literature Reviewneflibata-feng/MyArxiv-Agent12620 repos~5.9kAutomated safety check: NotesMIT
Preprint Search on bioRxivLigphiDonk/Oh-my--paper73912 repos~3.7kAutomated safety check: PassMIT
Academic Paper Writing PipelineImbad0202/academic-research-skills51k—~16kAutomated safety check: PassCustom licence

Similar skills

  • Nature Paper Card

    Yuan1z0825/nature-skills

    Builds a structured deep-reading card for one scientific paper, covering methods, how experiments support claims, limitations and research ideas, with a script to prepare the source.

    47k GitHub starsUsed in 2 repos~2.1k tokens
    Research & ScienceAuto-check passed
  • Systematic Review Screener

    Imbad0202/academic-research-skills

    Screens records for systematic, scoping and rapid reviews against fixed eligibility rules, using two blinded AI reviewers and a third adjudicator, with traceable PRISMA counts.

    51k GitHub stars~8.4k tokensUpdated yesterday
    Research & ScienceAuto-check passed
  • Literature Review

    neflibata-feng/MyArxiv-Agent

    Conduct comprehensive, systematic literature reviews using multiple academic databases (PubMed, arXiv, bioRxiv, Semantic Scholar, etc.).

    126 GitHub starsUsed in 20 repos~5.9k tokens
    Research & ScienceAuto-check: notes
  • Preprint Search on bioRxiv

    LigphiDonk/Oh-my--paper

    Searches bioRxiv life sciences preprints by keyword, author, date range or category with a Python script, returning JSON metadata and optional PDF downloads.

    739 GitHub starsUsed in 12 repos~3.7k tokens
    Research & ScienceAuto-check passed
  • Academic Paper Writing Pipeline

    Imbad0202/academic-research-skills

    Runs a 12-agent pipeline that plans, drafts, cites, reviews and formats academic papers, with modes for revision, rebuttals, abstracts and citation checks.

    51k GitHub stars~16k tokensUpdated yesterday
    Research & ScienceAuto-check passed
  • Deep Research Workflow

    TokenRhythm/opensquilla

    Runs multi-round research in three stages with a persisted state file, evidence tracking and a long-form report with per-claim citations.

    7.1k GitHub stars~1.3k tokensUpdated yesterday
    Research & ScienceAuto-check passed

More from huangwb8/ChineseResearchLaTeX

All 27 skills in this repo
  • NSFC Grant Rationale Writer

    huangwb8/ChineseResearchLaTeX

    Writes, restructures, reviews and polishes the rationale section of NSFC research grant applications in LaTeX, with backups and a diff before every write.

    2.9k GitHub stars~945 tokensUpdated 7 days ago
    Auto-check passed
  • LaTeX Example Content Generator

    huangwb8/ChineseResearchLaTeX

    Fills an existing LaTeX project with sample sections, tables and figure narratives, protecting the template structure and previewing before writing.

    2.9k GitHub starsUsed in 1 repo~744 tokens
    Auto-check passed
  • Journal Selector for Manuscripts

    huangwb8/ChineseResearchLaTeX

    Recommends journals for a manuscript by filtering a bundled impact-factor catalog, verifying scope and quality online, and writing a ranked Markdown report.

    2.9k GitHub stars~1.7k tokensUpdated 7 days ago
    Auto-check passed
  • NSFC Abstract Writer

    huangwb8/ChineseResearchLaTeX

    Writes Chinese and English abstracts for NSFC grant applications, with a recommended title and five alternatives, within set character limits.

    2.9k GitHub starsUsed in 1 repo~1.3k tokens
    Auto-check passed
  • NSFC Budget Justification Writer

    huangwb8/ChineseResearchLaTeX

    Writes a submission-ready NSFC budget justification as a LaTeX project and renders budget.pdf from your grant proposal text and supporting materials.

    2.9k GitHub starsUsed in 1 repo~1.4k tokens
    Auto-check passed
  • NSFC Application Code Recommender

    huangwb8/ChineseResearchLaTeX

    Recommends five pairs of NSFC application codes, primary and secondary, from a grant proposal's text and writes the reasons to a Markdown report.

    2.9k GitHub starsUsed in 1 repo~1.1k tokens
    Auto-check passed

Questions about Research Literature Radar

What does Research Literature Radar do?

发现、筛选并长期归档重要研究论文;当用户要求按主题寻找经典、rising star、社区精选、热点或顶会/顶刊论文时使用。采用分层发现策略,调用 research-literature-search 完成其中的关键词/数据库检索,再由本 skill 负责多渠道汇总、idea-level 价值判断、论文类型分类、跨轮次跟踪和论文库归档。. Research Literature Radar is an agent skill from huangwb8/ChineseResearchLaTeX.

When should I use Research Literature Radar?

Research Literature Radar fits situations like: tasks that involve Literature review.

How do I install Research Literature Radar in Claude Code?

Run `npx skills add huangwb8/ChineseResearchLaTeX --skill research-literature-radar -a claude-code`. Or copy the skill folder (skills/research-literature-radar in huangwb8/ChineseResearchLaTeX) into .claude/skills/research-literature-radar in your project. Claude Code loads it when a task matches its description.

How do I install Research Literature Radar in Codex?

Run `npx skills add huangwb8/ChineseResearchLaTeX --skill research-literature-radar -a codex`. Or copy the skill folder (skills/research-literature-radar in huangwb8/ChineseResearchLaTeX) into .agents/skills/research-literature-radar in your project. Codex loads it when a task matches its description.

Can I use Research Literature Radar 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 huangwb8/ChineseResearchLaTeX --skill research-literature-radar -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/research-literature-radar, .gemini/skills/research-literature-radar, .github/skills/research-literature-radar and .opencode/skills/research-literature-radar in your project.

What does Research Literature Radar need to run?

Going by SKILL.md and its folder, Research Literature Radar needs Python for the scripts in its folder and the command-line tools its instructions call (gh). Our summary lists: Python 3.

Does Research Literature Radar access the network?

SKILL.md contains no URLs. Its commands use gh, which can reach the network depending on how they are called. This is read from the text; nothing was executed.

Is Research Literature Radar 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 Research Literature Radar use?

Research Literature Radar is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Research Literature Radar use?

About 1.5k tokens (SKILL.md is roughly 6.2k 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 Research Literature Radar?

Skills that share tags, products or a category with Research Literature Radar: Nature Paper Card (Yuan1z0825/nature-skills, 47k stars), Systematic Review Screener (Imbad0202/academic-research-skills, 51k stars), Literature Review (neflibata-feng/MyArxiv-Agent, 126 stars) and Preprint Search on bioRxiv (LigphiDonk/Oh-my--paper, 739 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Research Literature Radar?

huangwb8 (a GitHub user) maintains it in huangwb8/ChineseResearchLaTeX, which has 2,880 GitHub stars. The repository holds 27 skills in this directory. The repository was last updated on October 4, 2026.

Source: huangwb8/ChineseResearchLaTeX on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.