canxiangcc/aminer-open-skill

60 stars Качество 40 Тренд 40

Official Research Skills Powered By AMiner.cn

Обзор

An AMiner Skill collection for finding papers, building reading lists and taxonomies, tracing sources, and checking citations in Claude Code, Codex, OpenClaw, and similar AI assistants. A typical workflow moves from finding literature, to understanding a paper's sources, to verifying its citations. Start with the category that matches your current task. Add the selected skills/ / directory to your AI assistant using its normal Skill or plugin installation method. If your assistant supports Claude plugins, use .claude-plugin/marketplace.json as the plugin list. Generate a token in the AMiner Console, then set AMINER_API_KEY in the environment used by your assistant: For local Claude Code, Codex, and similar conversational sessions, the repository provides quick setup tools: OpenClaw uses its own environment configuration: Standalone CLI commands, CI jobs, and scheduled tasks must set the token where the command runs.

README

aminer-open-skill

English | 中文

An AMiner Skill collection for finding papers, building reading lists and taxonomies, tracing sources, and checking citations in Claude Code, Codex, OpenClaw, and similar AI assistants.

🧰 Choose a Skill
🚀 Quick Start
💬 Use Cases
ℹ️ Notes
📚 References

Choose a Skill

A typical workflow moves from finding literature, to understanding a paper’s sources, to verifying its citations. Start with the category that matches your current task.

🔎 Find and collect literature

Skill	Use it when you want to	Token	Guide
`aminer-free-academic`	Find and screen papers, scholars, institutions, venues, or patents with free AMiner APIs	A token is still required; listed APIs are free	SKILL.md
`aminer-academic-search`	Run full academic searches, deeper analysis, or explicit structured experiment retrieval across papers, scholars, institutions, venues, and patents	Required; some APIs are billed; experiment search is ¥0.10/call	SKILL.md
`aminer-daily-paper`	Get personalized paper recommendations from topics, scholars, authors, or an AMiner account	Required	SKILL.md
`aminer-deep-search`	Build a large survey bibliography through multi-round search, deduplication, and citation expansion	Required	SKILL.md
`deep-research`	Produce a cited, hierarchically numbered research report plus a reusable Evidence Ledger from AMiner data and the host’s web tools	Required; free-first, billed APIs capped (¥20 hard stop)	SKILL.md

🗂️ Organize paper collections

Skill	Use it when you want to	Token	Guide
`scientific-taxonomy`	Build a hierarchical Markdown taxonomy from supplied paper titles and abstracts	Optional; only needed when AMiner must fill missing metadata	SKILL.md

🧭 Analyze and trace a paper

Skill	Use it when you want to	Token	Guide
`paper-source-trace`	Trace a paper’s key claims to its citation contexts and sources, then generate evidence reports and citation maps	Optional; only needed for AMiner enrichment	SKILL.md / Usage guide
`aminer-pdf-ocr`	Convert a paper PDF to Markdown via the AMiner PDF OCR service, then have the Agent extract experiment methods, datasets, and metrics with the built-in prompt	Required	SKILL.md
`aminer-exp-extraction`	Extract structured experiment JSON (methods, datasets, metrics, key results) from paper Markdown via a deterministic two-stage GLM pipeline (sentence filter + extraction); if you only have a PDF, run `aminer-pdf-ocr` first to get the Markdown	Requires `BIGMODEL_API_KEY` (Zhipu BigModel); no AMiner token needed	SKILL.md

✅ Verify citations

Skill	Use it when you want to	Token	Guide
`pdf-citation-verifier`	Check whether the references listed in a paper PDF actually exist	Required	SKILL.md
`structured-reference-audit`	Build a GROBID-backed reference ledger, then conservatively resolve each parsed entry through AMiner	Required for AMiner resolution; GROBID or TEI required	SKILL.md
`citation-faithfulness`	Check whether in-text citations accurately represent what the cited sources say	No AMiner token required; web access is required	SKILL.md

Quick Start

1. 📦 Add the Skill you need

Clone the repository:

git clone https://github.com/CanXiangCC/aminer-open-skill.git
cd aminer-open-skill

Add the selected skills// directory to your AI assistant using its normal Skill or plugin installation method. If your assistant supports Claude plugins, use .claude-plugin/marketplace.json as the plugin list.

2. 🔑 Configure the token when required

Generate a token in the AMiner Console, then set AMINER_API_KEY in the environment used by your assistant:

export AMINER_API_KEY=""

For local Claude Code, Codex, and similar conversational sessions, the repository provides quick setup tools:

Windows:       tools\setup-aminer-token.cmd
PowerShell:    .\tools\setup-aminer-token.ps1
macOS/Linux:   ./tools/setup-aminer-token.sh

OpenClaw uses its own environment configuration:

openclaw config set env.vars.AMINER_API_KEY ""

Standalone CLI commands, CI jobs, and scheduled tasks must set the token where the command runs. Check the table above first because not every Skill requires a token.

3. 💬 Ask naturally

Describe the academic task and provide any required paper, topic, scholar, or output preferences:

Find recent papers on multimodal agents and summarize the main research directions.

Skills that provide slash commands can also be invoked directly:

/aminer-deep-search topic: "multimodal agents" target-size: 200

Use Cases

The following prompts can be used directly after the corresponding Skill is installed:

Goal	Example request	Skill
Find papers quickly	“Find 10 recent papers on long-context language models. Return the title, year, venue, citation count, and URL.”	`aminer-free-academic`
Investigate a scholar or research topic	“Build a research profile for Andrew Ng, including interests, representative papers, collaborators, and recent work.”	`aminer-academic-search`
Retrieve structured experiments	“Return the original Experiment JSON for this paper, filtered by method and dataset name.”	`aminer-academic-search`
Get a focused reading list	“Recommend 8 papers on tool-using multimodal agents, prioritizing recent and highly cited work.”	`aminer-daily-paper`
Collect literature for a survey	“Collect 200 candidate papers on retrieval-augmented generation, expand from strong seed papers, remove duplicates, and export the bibliography.”	`aminer-deep-search`
Organize a paper collection	“Group these paper titles and abstracts into a coherent hierarchical taxonomy and save it as taxonomy.md.”	`scientific-taxonomy`
Produce a sourced research report	“Research the Chinese LLM industry and write a cited report with an evidence ledger backing every claim.”	`deep-research`
Trace a paper’s sources	“Analyze this PDF. Trace each key claim to its citation context and source, explain the citation intent, and generate the Markdown, JSON, SVG, and HTML outputs.”	`paper-source-trace`
Parse a PDF and extract experiments	“OCR this paper PDF, convert it to Markdown, and extract the experiment methods, datasets, and metrics as JSON.”	`aminer-pdf-ocr`
Extract experiments from Markdown	“Read this paper’s Markdown file and extract the experiment methods, datasets, metrics, and key results as one JSON file.”	`aminer-exp-extraction`
Detect fabricated references	“Check every reference in this paper PDF and flag entries that are missing, suspicious, or need manual review.”	`pdf-citation-verifier`
Audit parsed bibliography entries conservatively	“Parse this PDF into a reference ledger, then resolve each eligible reference through AMiner without treating an unresolved match as fabrication.”	`structured-reference-audit`
Check citation faithfulness	“For each in-text citation in this paper, retrieve the cited source and determine whether the surrounding claim is supported by the original text.”	`citation-faithfulness`

Notes

Never print, log, or commit AMINER_API_KEY.
aminer-exp-extraction calls the Zhipu BigModel API instead: set BIGMODEL_API_KEY for it (OPENAI_API_KEY accepted as fallback). It never uses the AMiner token.
Free AMiner APIs still require a token. Start with aminer-free-academic before using paid APIs.
Some tasks may incur API charges. Review the estimated cost before approving large searches or expensive calls.
The setup tools under tools/ configure local chat sessions only. Set the token separately for OpenClaw, CLI, CI, and scheduled jobs.
For direct API integration rather than Skill usage, see the AMiner Open Platform documentation below.

References

View this README on GitHub

Установка

npx skillfish add canxiangcc/aminer-open-skill