English | 中文

A growing knowledge infrastructure for academic research — not just another AI skill.

The market is flooded with AI skills for academic writing. Most do one thing: polish a sentence, format a citation, generate an outline. They're useful in isolation, but they forget everything between sessions and can't learn from each other.

This framework is the root system those skills plug into.

Most AI research tools are leaves — isolated, stateless, and narrow. This framework is the root system:

• It absorbs external skills (instead of competing with them)
• It learns from papers through systematic deconstruction
• It monitors field trends via social media and preprint channels
• Everything it learns accumulates into a structured, traceable, AI-callable knowledge base

The result compounds. Each paper you dissect makes the next one easier to write. Each skill you integrate becomes part of the shared foundation.

07_Scouting/external_skill_repos/

As new AI academic skills emerge on GitHub, this framework doesn't replace them — it absorbs them. Skills are smoke-tested, mapped to framework modules, and integrated as callable components:

The framework grows stronger as the skill ecosystem grows, rather than competing with it.

08_Case_Deconstruction/ + journal_access_lab/

Papers are the densest source of field-specific knowledge. This framework extracts that knowledge systematically:

Paper HTML/PDF
    ↓
Data-First Packet  ← What datasets? What variables? How constructed?
    ↓
Story Pattern      ← How is the problem framed? What's the gap logic?
    ↓
Method Card        ← What method, why, what are the robustness checks?
    ↓
Figure Pattern     ← What chart type, what it proves, how styled?
    ↓
Expression Bank    ← How does this journal write results like these?
    ↓
Backfill → 01_Data / 02_Story / 03_Methods / 04_Visualization / 05_Writing

Current scale (urban studies example):

• 500 papers processed for figure extraction (3,800+ high-res figures indexed)
• 60 papers fully acquired (A1 priority set)
• 15 gold-refined dissections with full traceability
• 45 standardized data-first packets
• 25 journals covered (Nature Cities → regional journals)

07_Scouting/social_public_platforms/

Research trends surface on social platforms before they reach journals. This module monitors:

• Creator watchlists — researchers who consistently share emerging work
• Platform watchlists — channels where field-specific insights aggregate
• Case candidate queues — papers worth dissecting flagged from social signals
• Comment/issue mapping — identifying methodology questions the field hasn't resolved

This feeds the 08_Case_Deconstruction/ pipeline with a continuously updated target list.

AI-Auto-Paper-Framework/
├── 00_Governance/       Project charter · Roadmap · Interactive dashboard
├── 01_Data/             Data source cards · Variable construction · Reliability ratings
├── 02_Story/            Problem framing · Gap logic · Mechanism chains · Contribution templates
├── 03_Methods/          20+ method cards (DID · Spatial Durbin · GWR · PSM · PCA · LDA…)
├── 04_Visualization/    Figure type index · R/Python templates · 500-paper figure library
├── 05_Writing/          4-tier writing patterns · Sentence banks · Journal-specific style guides
├── 06_Standards/        25 journal templates with submission checklists
├── 07_Scouting/         External skill registry · Social platform monitors
├── 08_Case_Deconstruction/  Paper packets · Pattern matrices · Gold refined set
├── 09_Templates/        Standardized intake forms
└── 10_Troubleshooting/  Known issues and solutions

journal_access_lab/      Paper acquisition automation
├── scripts/             60+ Node.js & PowerShell scripts (Edge CDP-based)
└── inventories/         2018–2026 article metadata (25 journals × 9 years)

The urban studies content is example scaffolding, not a constraint. The framework is designed to be swapped:

1. Replace 06_Standards/ with your target journals
2. Replace 08_Case_Deconstruction/ packets with papers from your field
3. Replace 01_Data/ source cards with your discipline's datasets
4. Replace 07_Scouting/ watchlists with relevant researchers and platforms
5. Keep 02_Story/, 03_Methods/, 05_Writing/ — these are largely universal

Fields this structure maps to cleanly: ecology, economics, public health, materials science, computational social science, climate science, urban planning.

git clone https://github.com/lightfor123/ai-auto-paper-framework.git
cd ai-auto-paper-framework

For the knowledge modules (AI-Auto-Paper-Framework/): no setup required. Browse and adapt the Markdown templates directly.

For the acquisition scripts (journal_access_lab/):

# Configure your local paper storage path
cp journal_access_lab/config.local.example.json journal_access_lab/config.local.json
# Edit config.local.json → set "paperLibraryPath"

# Or use an environment variable
export PAPER_LIBRARY_PATH=/path/to/your/paper/library

Scripts use Node.js built-ins; no npm install needed for most operations. Edge CDP-based scripts require Microsoft Edge.

MIT — use it, fork it, adapt it to your field.