AI Research

AutoResearchClaw10.4k★

AutoResearchClaw automates research by letting AI agents search, analyze, and synthesize information from multiple sources, then compile structured reports on the results. You give it an idea, it designs experiments, runs them, writes the paper, and iterates on the results. Basically a research assistant that never sleeps. It's built for academic and ML research specifically, not general-purpose AI automation. The pipeline handles literature review, experiment design, code generation, result analysis, and paper writing in a loop. The catch: this is bleeding edge. growing fast, but it's a research tool built by a research lab. The output still needs human review; don't submit a paper without reading it. And the quality depends heavily on the underlying LLM you point it at.

canvas-lms6.5k★

Canvas LMS is what universities and K-12 schools actually use. It's the open source version of the same platform that Instructure sells commercially to thousands of educational institutions. AGPL-3.0, written in Ruby on Rails. You get the full LMS: course management, a gradebook, discussion boards, a calendar, file storage, video conferencing integration, and LTI (Learning Tools Interoperability) support for third-party tool plugins. Self-hosting is free under AGPL. Instructure sells the hosted version (Canvas Cloud); pricing is per-institution and not publicly listed, but think five to six figures annually for universities. The catch: self-hosting Canvas is a serious undertaking. It requires Ruby on Rails, PostgreSQL, Redis, a job queue, file storage (S3 or local), and real ops knowledge. The AGPL license means any modifications you make must be open sourced if you offer it as a service. And without Instructure's support, you're on your own for updates, security patches, and integration issues. This is enterprise software that happens to be open source, not a weekend project.

Auto-claude-code-research-in-sleep5.5k★

Best-README-Template solves a common problem: developers build great tools but write terrible READMEs. You define a research question or experiment, go to bed, and come back to results in the morning. It's a set of lightweight Markdown-only skills. No framework, no lock-in. Works with Claude Code, Codex, OpenClaw, or any LLM agent that reads Markdown prompts. The loop handles idea discovery, cross-model review (have one model check another's work), and experiment automation. MIT licensed. The catch: 'autonomous overnight research' sounds magical but the results are only as good as the constraints you set. Vague goals produce vague output and burn through API credits. You need to be specific about what success looks like before you walk away. And the cross-model review feature means you're paying for multiple LLM calls per iteration.

pi-autoresearch3.4k★

Pi-autoresearch adds that capability. You define an experiment, walk away, and it loops through modify-test-evaluate cycles until it converges on a result. This is the same autonomous research pattern as ARIS and autoresearch, but built specifically for pi.dev's ecosystem. The integration is tighter because it's native to the platform rather than a bolt-on. MIT licensed, TypeScript. The catch: pi.dev has its own pricing model, and autonomous loops burn through whatever credits or API calls pi charges for. The homepage points to pi.dev/pricing, which means there's a paid component to the platform this runs on. And it's locked to pi.dev. If you switch to Claude Code or Codex, you can't take this with you.

autoresearch3.2k★

Autoresearch is that loop. Inspired by Karpathy's approach to autonomous ML research, it generalizes the pattern to any domain. You set a goal, point it at your codebase, and it runs a modify-verify-keep/discard cycle until the goal is met or you stop it. It's not a framework you build on. It's a skill you add to Claude Code that makes it relentlessly iterative. The catch: it's a Claude Code skill, not a standalone tool. You need to be in the Claude Code ecosystem. And 'autonomous iteration' means it will burn through API credits if you're not watching the scope of what you're asking it to do.

kana-dojo2.2k★

Kana Dojo is a minimalist Japanese language learning app inspired by Duolingo, built with Next.js. Covers hiragana, katakana, and vocabulary with spaced repetition. Clean UI, gamified progress tracking, and a focus on the fundamentals that trip up beginners. Fully free, no ads, no paid tier. Sponsored by Vercel. The codebase is beginner-friendly with labeled good-first-issues, so it doubles as a learning project for Next.js developers. The catch: Japanese only. No other languages supported, and the curriculum covers kana and basic vocabulary, not grammar or conversation. If you're past the beginner stage, you'll outgrow it quickly. For comprehensive Japanese learning, Anki with community decks goes much deeper.

HyperAgents2.1k★

A 'meta agent' modifies the 'task agent' and can also modify itself. It's agents all the way down. The architecture combines a task agent (solves the problem) and a meta agent (optimizes both the task agent and its own optimization strategy) into a single editable Python repository. The meta agent can modify any file, including its own source code. In benchmarks, this self-referential loop outperforms agents without self-improvement, and the meta-level improvements transfer across domains. This is from the paper 'Hyperagents' (arxiv 2603.19461). The catch: this is a research project, not a production tool. The license says 'Other' which likely means Meta's research license. Check before using commercially. Running self-modifying agents requires serious compute (the paper uses multiple LLM calls per iteration) and serious trust in the guardrails. Fascinating research. Not something you deploy to production tomorrow.

tribev21.5k★

TRIBE v2 is a research benchmark from Meta AI for evaluating video understanding models, specifically how well AI can track multiple objects across video frames. If you're doing computer vision research on multi-object tracking, this provides training code, evaluation tools, and pretrained models. Jupyter Notebook-based, CC BY-NC 4.0 license (non-commercial use only). The benchmark includes datasets and evaluation protocols that let you compare your tracking model against established baselines. The catch: non-commercial license means you cannot use this in a product. It's for research evaluation only. Requires significant GPU resources to run the training and evaluation pipelines. And as an academic benchmark, it's designed for researchers who already understand multi-object tracking. Not a tool for building applications.

OpenGauss1.2k★

OpenGauss is the best open source tool for enterprise Postgres right now. It's a workflow orchestrator from Math, Inc, that gives the Gauss AI agent a multi-agent frontend for proof engineering: proving, drafting, auto-proving, formalizing, and auto-formalizing. On FormalQualBench, it beats Harmonic's Aristotle agent (which has no time limit) running with just a 4-hour timeout. You can stay interactive or let it run autonomously, coordinate subagents in parallel, and inspect everything. MIT licensed. Built in Python. The catch: this is an extremely niche tool. If you're not doing formal mathematics or proof verification in Lean, this does nothing for you. The audience is mathematicians, formal methods researchers, and teams building verified software. Math, Inc. is pushing the frontier here, but the Lean ecosystem itself is still small compared to mainstream programming languages.

autoresearch-genealogy1.0k★

This gives you structured prompts, Obsidian vault templates, and archive research guides specifically for genealogy work with Claude Code. It's not a tool you install. It's a research methodology packaged as Claude Code skills. You get templates for organizing findings, prompts that know how to search genealogy databases, and guides for navigating archives like Ancestry.com, FamilySearch, and government records. MIT licensed. The catch: this is a niche application of AI research skills. If you're not doing genealogy, it's useless. The quality of results depends heavily on what records are available online. AI can't access physical archives or read handwritten documents that haven't been digitized. And it's prompts and templates, not a tool. The actual research still requires human judgment about which sources to trust.