update_readme_template.md

Awesome GUI Agent Paper List

This repo covers a variety of papers related to GUI Agents, such as:

• Datasets
• Benchmarks
• Models
• Agent frameworks
• Vision, language, multimodal foundation models (with explicit support for GUI)
• Works in general domains extensively used by GUI Agents (e.g., SoM prompting)

Papers Grouped by Environments

Web	Mobile	Desktop	GUI	Misc

(Misc: Papers for general topics that have important applications in GUI agents.)

Papers Grouped by Keywords

{{insert_keyword_groups_here}}

Papers Grouped by Authors

{{insert_author_groups_here}}

All Papers (from most recent to oldest)

Papers

{{insert_all_papers_here}}

How to Add a Paper or Update the README

Please fork and update:

• paper list
• README template
• automatic workflow

🤖 You can use this GPTs to quickly search and get a formatted paper entry automatically by inputting a paper name. Or you can simply leave a comment in an issue.

Format example and explanation

- [title](paper link)
    - List authors directly without a "key" identifier (e.g., author1, author2)
    - 🏛️ Institutions: List the institutions concisely, using abbreviations (e.g., university names, like OSU).
    - 📅 Date: e.g., Oct 30, 2024
    - 📑 Publisher: ICLR 2025
    - 💻 Env: Indicate the research environment within brackets, such as [Web], [Mobile], or [Desktop]. Use [GUI] if the research spans multiple environments. Use [Misc] if it is researching in general domains.
    - 🔑 Key: Label each keyword within brackets, e.g., [model], [framework],[dataset],[benchmark].
    - 📖 TLDR: Brief summary of the paper.

Regarding the 🔑 Key:

Key	Definition
model	Indicates a newly trained model.
framework	If the paper proposes a new agent framework.
dataset	If a new (training) dataset is created and published.
benchmark	If a new benchmark is established (also add "dataset" if there's a new training set).
primary studies	List the main focus or innovation in the study.
Abbreviations	Include commonly used abbreviations associated with the paper (model names, framework names, etc.).

For missing information, use "Unknown."