ZhangHanDong

ZhangHanDong / research-rs

Public

research cli for agent

35
2
100% credibility
Found Apr 21, 2026 at 35 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Rust
AI Summary

A command-line tool for managing reproducible research sessions by fetching online and local sources, building figure-rich reports, and maintaining a persistent wiki knowledge base.

How It Works

1
💡 Discover Research Helper

You hear about a friendly tool that keeps your research organized, never forgetting sources or ideas.

2
📝 Start Your Project

Name your topic like 'AI Agents' and create a new space to collect thoughts and findings.

3
🔗 Gather Sources

Add web pages or your own files, and it checks them to keep only the good stuff.

4
✨ Let It Research Automatically

Watch as it reads sources, builds a wiki of key ideas, draws helpful diagrams, and writes a clear story.

5
❓ Ask Questions

Query your personal wiki anytime to get answers based on what you've gathered.

🎉 Share Your Report

Enjoy a beautiful, illustrated page with your overview, findings, diagrams, and sources all ready to share.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 35 to 35 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is research-rs?

research-rs (now ascent-research) is a Rust CLI that powers incremental research sessions for AI agents, turning one-off LLM chats into durable workflows. Point it at a topic like "tokio internals," add sources via URLs or local code trees, and run autonomous loops with Claude or Codex to fetch content, build wiki pages, draw SVG diagrams, and generate figure-rich HTML reports. Sessions persist as plain files—Markdown, JSONL logs, SVGs—resumable across days, so agents accrete knowledge without resetting.

Why is it gaining traction?

It evolves karpathy's autoresearch loop for open-ended research, enforcing figure-rich outputs and coverage checks like unused sources or unresolved diagrams, unlike one-shot summarizers. Developers dig the hybrid ingest (HTTP APIs, browser fallback, local files) feeding a growing wiki you can query or lint, plus seamless integration as a skill in Claude Code setups. For github research repos or tools like research github copilot, it stands out by making agent memory inspectable via grep or Obsidian.

Who should use this?

AI agent builders embedding research into coding workflows, like diving into library source trees or surveying state-space models from arXiv and GitHub. Rust devs auditing crates via add-local on src dirs, or note-takers structuring Obsidian vaults into queryable wikis. Ideal for github research groups prototyping climate change analyses or clinical research over mixed online/local data.

Verdict

Try it for agent-driven github research blogs or data dives—solid for early experiments despite 35 stars and 1.0% credibility signaling raw maturity. Polish the docs and add more presets to hit escape velocity; pair with postagent for best results.

(187 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.