Agent skill

Input Router

by gaotiexinqu in gaotiexinqu/OneResearchClaw

Use the input path to select the correct downstream grounding pipeline and continue execution until the selected grounding workflow is completed.

MITAuto-check passedResearch & Science

Install Input Router

skills CLI
$ npx skills add gaotiexinqu/OneResearchClaw --skill input-router -a claude-code

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

GitHub CLI
$ gh skill install gaotiexinqu/OneResearchClaw input-router --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/gaotiexinqu/OneResearchClaw.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.cursor/skills/input-router .claude/skills/input-router && 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
input-router
GitHub stars
450
Token cost
~2.2k tokens
SKILL.md length
985 words
Files
1
Skills in repo
15
Repo updated
First seen
Licence
MIT

At a glance

Use the input path to select the correct downstream grounding pipeline and continue execution until the selected grounding workflow is completed.

  • Works in 5 steps: inspect the input path, → determine the file type strictly by… → select the correct downstream grounding… → …
  • Research & Science work in your project
  • SKILL.md covers What This Skill Does, What This Skill Does Not Do, Routing Principle and Supported Routing Table, plus 7 more sections
  • Reaches arxiv.org and youtube.com

What it does

Input Router is an agent skill from gaotiexinqu/OneResearchClaw. Use the input path to select the correct downstream grounding pipeline and continue execution until the selected grounding workflow is completed.

Its SKILL.md is about 2.2k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Research & Science. It works with arXiv and YouTube. The repository describes itself as: Any research. One Claw. 🦞 From any materials to research with fully autonomous & skill-driven researcher. The licence is MIT.

When your agent uses it

  • Research & Science work in your project

Example prompts

  • “/input-router”

Workflow steps

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

  1. inspect the input path,
  2. determine the file type strictly by extension,
  3. select the correct downstream grounding pipeline,
  4. invoke the corresponding existing skill,
  5. and continue until the selected downstream skill's workflow has completed.

What it can do on your machine

Read from SKILL.md and the folder at commit 37e86c6. 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

    No scripts in the folder and no shell commands in SKILL.md.

    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:

    • arxiv.org
    • youtube.com
    • bilibili.com
    • youtu.be
    • b23.tv

    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

Input Router loads about 2.2k tokens when it runs. Until then it costs about 40 tokens; SKILL.md has 985 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~40
When it runs · the whole SKILL.md, loaded when a task matches
~2.2k

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); files beside SKILL.md are not scanned.

SKILL.md

The full file from gaotiexinqu/OneResearchClaw at commit 37e86c6, republished under its MIT licence (© gaotiexinqu). 985 words, ~2,203 tokens.

Download SKILL.mdSave it as .claude/skills/input-router/SKILL.md (or your agent's skills folder).
name
input-router
description
Use the input path to select the correct downstream grounding pipeline and continue execution until the selected grounding workflow is completed.

Input Router

Use this skill as the unified input entry layer for the system.

This skill is a system-level dispatcher. It does not implement document parsing, transcription, table extraction, PPT grounding, archive unpacking, or meeting summarization logic by itself.

However, it does not stop at routing analysis. After identifying the correct input type, it must immediately invoke the correct downstream skill and continue the task until that downstream grounding workflow is completed.

What This Skill Does

This skill must:

  1. inspect the input path,
  2. determine the file type strictly by extension,
  3. select the correct downstream grounding pipeline,
  4. invoke the corresponding existing skill,
  5. and continue until the selected downstream skill's workflow has completed.

What This Skill Does Not Do

This skill must not directly implement the grounding logic that belongs to downstream skills. It must not:

  • transcribe audio by itself
  • extract content from documents by itself
  • parse tables by itself
  • ground PPTX files by itself
  • unpack archives by itself
  • write grounded.md by bypassing the downstream skill

Those tasks belong to the selected downstream skill.

Routing Principle

Route inputs strictly by file extension or URL pattern.

URL Detection (Priority 1)

If the input is a URL instead of a local file:

URL PatternRoute ToOutput Type
arxiv.org/abs/ or arxiv.org/pdf/remote-input.pdf
youtube.com/watch, youtu.be, youtube.com/shortsremote-input.mp4/.mkv
bilibili.com/video, b23.tvremote-input.mp4/.mkv

URL routing workflow:

  1. Detect URL pattern
  2. Invoke remote-input skill to download
  3. Use returned local path for downstream routing
    • If merge_failed: true, use audio_path instead of path
  4. Continue with extension-based routing

Note on merge failure: When remote-input returns merge_failed: true, the actual file for downstream routing is the audio file (.webm etc.), which routes to meeting-audio-grounding instead of meeting-video-grounding.

File Extension Detection (Priority 2)

For local files, route strictly by file extension:

ExtensionRoute To
.mp3, .wav, .m4a, .webm, .aac, .oggmeeting-audio-grounding
.mp4, .mov, .mkvmeeting-video-grounding
.pdf, .docx, .md, .txtdocument-grounding
.xlsx, .csvtable-grounding
.pptxpptx-grounding
.ziparchive-grounding

Do not use filename semantics, directory names, or inferred task intent to override the extension-based mapping.

Supported Routing Table

URL Patterns (Remote Inputs)
URL PatternRoute ToLocal Output
arxiv.org/abs/ or arxiv.org/pdf/remote-inputPDF in remote/arxiv/
youtube.com/watch, youtu.be, youtube.com/shortsremote-inputVideo in remote/youtube/
bilibili.com/video, b23.tvremote-inputVideo in remote/bilibili/
File Extensions (Local Inputs)
ExtensionRoute To
.mp3, .wav, .m4a, .webm, .aac, .oggmeeting-audio-grounding
.mp4, .mov, .mkvmeeting-video-grounding
.pdf, .docx, .md, .txtdocument-grounding
.xlsx, .csvtable-grounding
.pptxpptx-grounding
.ziparchive-grounding

Required Workflow

When using this skill, you must follow this workflow:

  1. Detect input type:

    • If input contains URL patterns (arxiv.org, youtube.com, youtu.be, bilibili.com, b23.tv): goto URL handling
    • Otherwise: goto file handling
  2. URL handling:

    1. Match URL pattern to remote-input
    2. Invoke remote-input skill
    3. Receive returned local file path
    4. Use the local path for downstream routing
  3. File handling:

    1. Inspect the input path
    2. Determine the file extension
    3. Match the extension to the routing table
    4. Select the corresponding downstream skill
  4. Continue:

    1. Immediately invoke the selected downstream skill
    2. Follow the downstream skill's own workflow strictly
    3. Do not stop after only reporting the routing decision

URL Detection Patterns

arXiv URLs
  • Pattern: https?://(?:www\.)?arxiv\.org/(abs|pdf)/
  • Examples:
    • https://arxiv.org/abs/2301.07041
    • https://arxiv.org/pdf/2301.07041.pdf
    • http://arxiv.org/abs/2301.07041
YouTube URLs
  • Pattern: https?://(?:www\.)?youtube\.com/(watch|shorts)/ or https?://youtu\.be/
  • Examples:
    • https://www.youtube.com/watch?v=dQw4w9WgXcQ
    • https://youtu.be/dQw4w9WgXcQ
    • https://www.youtube.com/shorts/abc123
Bilibili URLs
  • Pattern: https?://(?:www\.)?bilibili\.com/video/ or https?://b23\.tv/
  • Examples:
    • https://bilibili.com/video/BV1xx411c7JZ
    • https://www.bilibili.com/video/av12345678
    • https://b23.tv/abc123

Merge Failure Handling

When downloading video from URL (YouTube or Bilibili), the downloader may produce separate video and audio files instead of a merged video file (merge failure). This happens when:

  • The video has separate video and audio streams that couldn't be merged
  • Common pattern: .mp4 file without audio + .webm audio file
Show full SKILL.md (386 more words)Show less
Detection and Recovery

After remote-input downloads video content:

  1. Check return values for merge_failed field:

    • merge_failed: false → Use path as normal (video with audio)
    • merge_failed: true → Use audio_path instead of path
  2. Reroute based on actual file type:

    • If audio_path is used → route to meeting-audio-grounding
    • If path is used → route based on extension (video → meeting-video-grounding)
Routing After Merge Failure

When merge_failed: true, the local file is typically a pure audio file (.webm, .mp3, etc.):

Returned File TypeRoute To
.mp3, .wav, .m4a, .webm, .aac, .oggmeeting-audio-grounding
.mp4, .mov, .mkv (with audio)meeting-video-grounding
Workflow Example
User: https://youtube.com/watch?v=xxx
           ↓
remote-input downloads video
           ↓
Returns: {
  "path": "video.mp4",           // video without audio
  "audio_path": "audio.webm",    // separate audio file
  "merge_failed": true
}
           ↓
Since merge_failed=true, use audio_path
           ↓
audio.webm extension → .webm
           ↓
Route to: meeting-audio-grounding

Completion Rule

The task is not complete after:

  • identifying the URL pattern or extension,
  • naming the downstream skill,
  • or reporting the selected pipeline.

The task is complete only after:

  • for URL inputs: remote-input has successfully downloaded the content
  • the selected downstream skill has been invoked,
  • its required workflow has been followed,
  • and the expected grounding result for that input type has been produced.

Unsupported Inputs

If the input is neither a supported URL nor a supported file extension:

  • report that the input is unsupported by the current router,
  • do not guess a pipeline,
  • do not force the file through an unrelated skill.

Dispatch Behavior

The router is an entry skill, not a stopping point.

For URL inputs:

  • do not stop after saying remote-input; run it
  • use the returned local path to continue routing

For audio inputs, do not stop after saying meeting-audio-grounding; run it For video inputs, do not stop after saying meeting-video-grounding; run it For document inputs, do not stop after saying document-grounding; run it For table inputs, do not stop after saying table-grounding; run it For PPT inputs, do not stop after saying pptx-grounding; run it For ZIP inputs, do not stop after saying archive-grounding; run it

Practical Meaning

From the user's perspective, this skill should behave like a unified input layer:

For URL inputs:

  • the user provides a URL (arxiv paper or YouTube video),
  • this skill downloads the content to data/raw_inputs/remote/,
  • the task continues through the appropriate grounding pipeline.

For local file inputs:

  • the user provides one input file,
  • this skill selects the correct downstream grounding pipeline,
  • and the task continues through that pipeline until grounding is actually completed.

The user should not need to manually re-enter another instruction just because the router has already identified the correct skill.

© gaotiexinqu, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in .cursor/skills/input-router of gaotiexinqu/OneResearchClaw.

Open the folder on GitHubat commit 37e86c6

Compare with similar skills

Input Router 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.

Input Router compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Input Router this skillgaotiexinqu/OneResearchClaw450—~2.2kAutomated safety check: PassMIT
Larksnap FetchAmbroseX/larksnap291—~923Automated safety check: PassApache-2.0
Superlearnraiyanyahya/Superlearn122—~6.2kAutomated safety check: PassMIT
Insane Searchfivetaku/gptaku-plugins-codex128—~5.6kAutomated safety check: PassMIT
Horizontal-Vertical Deep ResearchKKKKhazix/khazix-skills21k—~2.1kAutomated safety check: PassMIT
Huashu Reportalchaincyf/huashu-report447—~1.4kAutomated safety check: PassMIT

Similar skills

  • Larksnap Fetch

    AmbroseX/larksnap

    把飞书/Lark 文档或普通网页抓取并保存到本地,也能编辑用户有权限的飞书文档,并用已登录浏览器执行一次网页搜索。用户要求下载、导出、抓取、写入飞书文档,或联网搜索资料/参考链接时使用本技能,即使没有提到 larksnap。底层通过技能自带 daemon 桥接已登录的 larksnap 浏览器扩展;arXiv 使用独立脚本。

    291 GitHub stars~923 tokensUpdated 1 mo ago
    Research & ScienceAuto-check passed
  • Superlearn

    raiyanyahya/Superlearn

    Build an interactive learning board on any topic. An agent skill from raiyanyahya/Superlearn.

    122 GitHub stars~6.2k tokensUpdated 1 mo ago
    Research & ScienceAuto-check passed
  • Insane Search

    fivetaku/gptaku-plugins-codex

    Adaptive access for blocked websites — tries every method until one works.

    128 GitHub stars~5.6k tokensUpdated 1 mo ago
    Research & ScienceAuto-check passed
  • Horizontal-Vertical Deep Research

    KKKKhazix/khazix-skills

    Runs a two-axis deep research method on a product, company, concept or person: its full history over time, compared with peers today, delivered as a typeset PDF report.

    21k GitHub stars~2.1k tokensUpdated 9 days ago
    Research & ScienceAuto-check passed
  • Huashu Report

    alchaincyf/huashu-report

    做机构级研究报告与学术论文——行业报告、白皮书、年度调研、数据洞察、arXiv论文,也含16:9咨询deck型报告(一页一结论)。定结构、写发现、建图表视觉系统。规范提炼自2026年顶级机构报告实测。单篇文章、纯演示PPT不适用(走design/slides)。

    447 GitHub stars~1.4k tokensUpdated 26 days ago
    Research & ScienceAuto-check passed
  • Paper Analyzer

    zsyggg/paper-craft-skills

    将学术论文转化为深度HTML长文。6轮强制工作流、代码仓库搜索、公式渲染、Mermaid图表. An agent skill from zsyggg/paper-craft-skills.

    1.3k GitHub stars~1.9k tokensUpdated 4 mo ago
    Research & ScienceAuto-check passed

More from gaotiexinqu/OneResearchClaw

All 15 skills in this repo
  • Remote Input

    gaotiexinqu/OneResearchClaw

    Download remote content (arxiv papers, YouTube videos, Bilibili videos) to local storage and route to downstream grounding pipeline.

    450 GitHub stars~2.8k tokensUpdated 5 mo ago
    Auto-check passed
  • Grounded Research Lit

    gaotiexinqu/OneResearchClaw

    Run focused literature and web research from a grounded note.

    450 GitHub stars~11k tokensUpdated 5 mo ago
    Auto-check passed
  • Archive Grounding

    gaotiexinqu/OneResearchClaw

    Unpack a ZIP archive, inventory its files, run the corresponding child grounding skill for each supported child file, and then write a real archive-level grounded.md.

    450 GitHub stars~2.1k tokensUpdated 5 mo ago
    Auto-check: notes
  • Document Grounding

    gaotiexinqu/OneResearchClaw

    Convert a raw document into a structured grounding note for downstream research and summarization.

    450 GitHub stars~2k tokensUpdated 5 mo ago
    Auto-check passed
  • Meeting Audio Grounding

    gaotiexinqu/OneResearchClaw

    Convert a meeting audio file into a transcript bundle, then use meeting-grounding to produce structured meeting grounding outputs.

    450 GitHub stars~1.1k tokensUpdated 5 mo ago
    Auto-check passed
  • Meeting Video Grounding

    gaotiexinqu/OneResearchClaw

    Convert a meeting video into an audio-first transcript bundle, then use meeting-grounding to produce structured meeting grounding outputs.

    450 GitHub stars~1.2k tokensUpdated 5 mo ago
    Auto-check passed

Works with

Questions about Input Router

What does Input Router do?

Use the input path to select the correct downstream grounding pipeline and continue execution until the selected grounding workflow is completed. Input Router is an agent skill from gaotiexinqu/OneResearchClaw. Use the input path to select the correct downstream grounding pipeline and continue execution until the selected grounding workflow is completed.

When should I use Input Router?

Input Router fits situations like: research & Science work in your project.

How do I install Input Router in Claude Code?

Run `npx skills add gaotiexinqu/OneResearchClaw --skill input-router -a claude-code`. Or copy the skill folder (.cursor/skills/input-router in gaotiexinqu/OneResearchClaw) into .claude/skills/input-router in your project. Claude Code loads it when a task matches its description.

How do I install Input Router in Codex?

Run `npx skills add gaotiexinqu/OneResearchClaw --skill input-router -a codex`. Or copy the skill folder (.cursor/skills/input-router in gaotiexinqu/OneResearchClaw) into .agents/skills/input-router in your project. Codex loads it when a task matches its description.

Can I use Input Router 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 gaotiexinqu/OneResearchClaw --skill input-router -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/input-router, .gemini/skills/input-router, .github/skills/input-router and .opencode/skills/input-router in your project.

What does Input Router need to run?

SKILL.md names no scripts, command-line tools or credentials: Input Router is instructions for the agent only.

Does Input Router access the network?

SKILL.md names 5 domains. In commands or code: arxiv.org, youtube.com, bilibili.com, youtu.be and b23.tv; the agent is likely to contact these when it follows the instructions. This is read from the text; nothing was executed.

Is Input Router 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. Review the folder before installing.

What licence does Input Router use?

Input Router 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 Input Router use?

About 2.2k tokens (SKILL.md is roughly 8.8k 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 Input Router?

Skills that share tags, products or a category with Input Router: Larksnap Fetch (AmbroseX/larksnap, 291 stars), Superlearn (raiyanyahya/Superlearn, 122 stars), Insane Search (fivetaku/gptaku-plugins-codex, 128 stars) and Horizontal-Vertical Deep Research (KKKKhazix/khazix-skills, 21k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Input Router?

gaotiexinqu (a GitHub user) maintains it in gaotiexinqu/OneResearchClaw, which has 450 GitHub stars. The repository holds 15 skills in this directory. The repository was last updated on May 9, 2026.

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