Hacker News

Article URL: https://grohan.co/2026/02/10/ctoc/

Comments URL: https://news.ycombinator.com/item?id=46969042

Points: 1

# Comments: 1

Article URL: https://www.newyorker.com/magazine/2026/02/16/what-is-claude-anthropic-doesnt-know-either

Comments URL: https://news.ycombinator.com/item?id=46969030

Points: 1

# Comments: 0

Article URL: https://platformonomics.com/2026/02/a-warning-to-seattle-dont-become-the-next-cleveland/

Comments URL: https://news.ycombinator.com/item?id=46969017

Points: 2

# Comments: 0

Article URL: https://www.wsj.com/tech/ai/openai-executive-who-opposed-adult-mode-fired-for-sexual-discrimination-3159c61b

Comments URL: https://news.ycombinator.com/item?id=46968988

Points: 2

# Comments: 2

Article URL: https://github.com/alibaba-damo-academy/RynnBrain

Comments URL: https://news.ycombinator.com/item?id=46968964

Points: 1

# Comments: 0

Article URL: https://modo.is/b/multimeter-stand-with-storage

Comments URL: https://news.ycombinator.com/item?id=46968900

Points: 1

# Comments: 0

Article URL: https://github.com/openai/codex/issues/11189

Comments URL: https://news.ycombinator.com/item?id=46968891

Points: 4

# Comments: 1

I'm an ML scientist who was drowning in papers. Every research tool I tried locked me into their workflow, their sources, their extraction schemas. The real friction wasn't features—it was the interface. Why am I editing files and settings when I could just tell the agent what I want? How do I find new research thats relevant to ME and keep up to date? How do i have my agent actually LEARN about my work and my research paths?

So I built Thoth. Three design decisions that drove everything:

1. HOT-LOADING SKILLS Agents start minimal—no tools loaded. When you ask "find papers on attention mechanisms," the agent loads the paper-discovery skill, which attaches its MCP tools dynamically. When you switch to "analyze this citation network," it loads the citation skill. 10 bundled skills, users can create their own. This keeps context windows small and agents focused.

2. CHAT AS CONFIGURATION Every setting, every source, every extraction schema can be changed through conversation. "Add NeurIPS as a source" just works. "Change the extraction schema to include methodology sections" just works. No config files required (though power users can still edit them).

3. AUTOMATED SOURCE DISCOVERY Give the agent any URL. It uses Playwright + an LLM to auto-detect article elements and creates a working scraper. Zero configuration. If it gets something wrong, tell it in natural language and it iterates. Then tell the agent about a broad research idea you have and let it help you refine that topic, create a new research job, automatically add sources and run recurring updates. You always have the most up to date research for YOUR work not just the hot topic of the day.

Other stuff: - 64 MCP tools across 16 categories - Letta-powered persistent memory (6 memory blocks per agent) - Hybrid RAG: pgvector + BM25 + Reciprocal Rank Fusion + reranking - 6-stage citation resolution (Crossref → OpenAlex → ArXiv → fuzzy matching) - 7 pre-built source plugins (ArXiv, Semantic Scholar, NeurIPS, ICML, etc.) - One-command install: curl | bash - Obsidian plugin for the UI - Fully local / privacy-first

Stack: Python 3.12, FastAPI, Letta, PostgreSQL+pgvector, TypeScript, Docker

https://github.com/acertainKnight/project-thoth

Would love feedback. Defnitiely will have some bugs here and there but its been great for my own workflows.

Comments URL: https://news.ycombinator.com/item?id=46968889

Points: 1

# Comments: 0

Article URL: https://jonathannen.com/daily-ai-chat

Comments URL: https://news.ycombinator.com/item?id=46968884

Points: 1

# Comments: 0

Article URL: https://eruci.com/t.html

Comments URL: https://news.ycombinator.com/item?id=46968863

Points: 1

# Comments: 0

Article URL: https://runreplay.com

Comments URL: https://news.ycombinator.com/item?id=46968830

Points: 1

# Comments: 1

Working on an AI companion (https://t.me/adola2048_bot) with persistent memory across sessions. What approaches have others used for the memory layer?

Comments URL: https://news.ycombinator.com/item?id=46968827

Points: 1

# Comments: 0

Article URL: https://devblogs.microsoft.com/dotnet/dotnet-11-preview-1/

Comments URL: https://news.ycombinator.com/item?id=46968815

Points: 2

# Comments: 0

The decentralized web at its purest...

This is a fun toy project, but with some real-world use cases: simple domain parking, business card page, temporary maintenance page (but not suggested)

It operates as a proxy — a truly decentralized, DNS-like experience would require broad adoption or formal standardization

Comments URL: https://news.ycombinator.com/item?id=46968802

Points: 1

# Comments: 0

Article URL: https://www.cnn.com/2026/02/10/health/fda-moderna-mrna-flu-vaccine

Comments URL: https://news.ycombinator.com/item?id=46968800

Points: 5

# Comments: 0

Article URL: https://www.youtube.com/watch?v=Sxmd7T_dyaM

Comments URL: https://news.ycombinator.com/item?id=46968797

Points: 1

# Comments: 0

Article URL: https://www.mattgreer.dev/blog/gba-dev-drawing-pixels/

Comments URL: https://news.ycombinator.com/item?id=46968794

Points: 1

# Comments: 0

Article URL: https://airstack.garden/

Comments URL: https://news.ycombinator.com/item?id=46968788

Points: 1

# Comments: 0

Article URL: https://blog.fsck.com/2026/02/10/dorodango/

Comments URL: https://news.ycombinator.com/item?id=46968778

Points: 1

# Comments: 0

Article URL: https://github.com/flawnn/ArcFolderArchiver

Comments URL: https://news.ycombinator.com/item?id=46968319

Points: 1

# Comments: 0