Journal Recommender Skill

journal-recommender is an Agent Skill for recommending academic journals from a manuscript abstract. It is designed for graduate students and researchers who need a practical shortlist of journals for submission.

The skill combines abstract-based topic matching, current journal metadata, official Aims & Scope evidence, CAS/JCR ranking, review-speed and OA/APC considerations, and publication-risk checks.

It follows the open Agent Skills style and can be used in skills-compatible agent runtimes such as Claude Code, Codex, Cursor, OpenClaw, and similar tools.

What It Does

• Identifies the manuscript's main discipline and optional secondary disciplines.
• Estimates a tentative submission strategy from abstract signals.
• Builds a challenge / target / safety journal pool.
• Uses current web data instead of relying on memory for IF, rankings, APCs, indexing, or review time.
• Scores journals using evidence from LetPub or equivalent metadata pages, official journal pages, and Aims & Scope text.
• Flags publication risks such as warning-list status, ESCI-only indexing, high self-citation, unclear APCs, or incomplete verification.
• Produces a concise final recommendation, usually challenge 2 + target 2 + safety 1.

When To Use

Use this skill when a user asks things like:

• "推荐几个适合这篇论文投稿的期刊"
• "我的论文适合投什么期刊？"
• "帮我选刊"
• "Which journals should I submit this paper to?"
• "Find 3-5 suitable journals for this abstract"

Repository Structure

journal-recommender/
├── SKILL.md
└── references/
    ├── cas-subjects.md
    ├── data-sources.md
    ├── output-templates.md
    ├── risk-checks.md
    └── scoring-rubric.md

SKILL.md contains the core workflow and points the agent to the reference files only when needed. The reference files hold the longer scoring rules, data-source guidance, risk checks, and output templates.

Installation

Option 1: One-line install

If your agent runtime supports skill installation, ask the agent directly:

帮我安装这个 skill：https://github.com/zero565656/journal-recommender

Or use the general skills installer:

npx skills add zero565656/journal-recommender

If needed, specify the target agent/runtime:

npx skills add zero565656/journal-recommender -a codex
npx skills add zero565656/journal-recommender -a claude-code
npx skills add zero565656/journal-recommender -a cursor
npx skills add zero565656/journal-recommender -a openclaw

Option 2: Manual install

Clone this repository into the skills directory used by your agent runtime.

Runtime	Install path
Claude Code	~/.claude/skills/journal-recommender/
Codex CLI	~/.codex/skills/journal-recommender/
Cursor	~/.cursor/skills/journal-recommender/
OpenClaw	~/.openclaw/workspace/skills/journal-recommender/
Other skills-compatible agents	Clone into that runtime's skills/ directory

macOS / Linux:

git clone https://github.com/zero565656/journal-recommender.git ~/.codex/skills/journal-recommender

Windows PowerShell:

$skills = "$env:USERPROFILE\.codex\skills"
New-Item -ItemType Directory -Force $skills | Out-Null
git clone https://github.com/zero565656/journal-recommender.git "$skills\journal-recommender"

Replace the destination path if you are installing for Claude Code, Cursor, OpenClaw, or another runtime.

Restart your agent after installation so the skill can be discovered.

Option 3: Use as a reference

If your agent does not support Agent Skills auto-loading, open SKILL.md and paste its contents into your conversation as instructions. The skill is plain Markdown with YAML frontmatter plus reference files.

Updating

Go to the installation path for your runtime, then pull the latest version:

cd ~/.codex/skills/journal-recommender
git pull

On Windows PowerShell:

cd "$env:USERPROFILE\.codex\skills\journal-recommender"
git pull

Replace the path if you installed it under Claude Code, Cursor, OpenClaw, or another runtime.

Example Prompt

Use journal-recommender. Here is my abstract:

[Paste abstract here]

I prefer SCI journals, CAS 1-2区 if possible. My deadline is about 8 months, and APC should be under $3000.

Notes

• The skill requires current web access for reliable journal recommendations.
• It should not fabricate journal facts from memory.
• Abstract-only tiering is a submission-strategy estimate, not an objective judgment of manuscript quality.
• Final submission decisions should still be checked against supervisor advice, school graduation requirements, and the journal's official website.