bettyguo/awesome-research-agents — explained in plain English
Analysis updated 2026-05-18
Find vetted MCP servers and agents for literature search, reading, and experiment tracking.
Compare tools like Elicit and AI2 Scholar QA before choosing one for a literature review.
Pick a PDF to Markdown tool such as Marker or GROBID for processing scientific papers.
| bettyguo/awesome-research-agents | adguardteam/ruleseditor | baiyuetribe/serverstatus-theme | |
|---|---|---|---|
| Stars | 7 | 7 | 7 |
| Language | — | TypeScript | CSS |
| Last pushed | — | 2026-07-01 | 2019-07-21 |
| Maintenance | — | Active | Dormant |
| Setup difficulty | easy | easy | moderate |
| Complexity | 1/5 | 2/5 | 2/5 |
| Audience | researcher | developer | ops devops |
Figures from each repo's GitHub metadata at analysis time.
Awesome Research Agents 2026 is a curated list built for machine learning researchers, not another giant directory of every AI tool that exists. The author reviewed the space and picked 44 entries, organized by the stages of a real research workflow: discovering papers, reading and annotating them, running experiments, writing code, evaluating results, writing papers, sharing work, and staying current with new releases. That structure matters because a junior researcher usually knows what task they need to finish, but not whether the right tool for it counts as an agent framework, a skill, or an MCP server. What sets this list apart from typical roundups is that every entry includes a short "Why we recommend it" note explaining something the tool's own documentation would not say, plus a one line "When not to use" warning. So instead of just listing options, the project states the tradeoffs plainly: which tools are actually maintained, which are thin wrappers around something else, which have a free tier worth using, and which only make sense in narrow situations. Examples from the README include Elicit for structured literature review, which is closed source but has a real free tier, AI2 Scholar QA for open, citation backed literature synthesis, small focused MCP servers for searching arXiv and Semantic Scholar, and Hugging Face's official MCP server for giving an AI agent live access to models and datasets. On the reading side it covers Zotero for reference management, a Better BibTeX plugin for stable citation keys, and PDF to Markdown tools such as Marker and GROBID for turning scientific papers into structured, searchable text. Entries are checked again every quarter. Tools that stop being maintained get cut, and stronger alternatives take their place. The project accepts pull requests but holds them to a high bar, since the entire point of the list is judgment, not coverage. This is a reference for researchers choosing tools, not a piece of software you install or run yourself.
A curated, opinionated list of 44 AI tools and MCP servers that machine learning researchers actually use, organized by research workflow stage.
Setup difficulty is rated easy, with roughly 5min to a first successful run.
Mainly researcher.
This repo across BitVibe Labs
Verify against the repo before relying on details.