Everybody seems to be talking about AI agents and filesystems these days, and for good reasons...📁 Giving your agent access to filesystem exploration and bash tools can significantly improve its information retrieval capacity and the context in which it lives⚡

New Rust crate: 𝗮𝗿𝘅𝗶𝘃-𝗰𝗹𝗶🦀 I built a lightweight CLI app that allows you to download the most recent papers belonging to a specific category on arXiv, with their metadata, plain-text summaries and PDF versions📁 You can download metadata and text summaries for 1000 papers in 0.5s!🚀

Why would you want one LLM provider for your coding agent when you can have many?🤖

I needed to upload JSONL data to an SQLite DB today: I wanted to do it with python, but then I thought: why don't I build this in Rust?🦀

I love the concept of sandbox⏳ Especially if it means that my coding agent gets to play with a virtual file system while all my files are safe from the next `rm -rf /`🔒

The US has openly declared that it can steal other countries and run them itself and the world should unite against this aggression.

These past few days I've been cooking🍳 Say hello to 𝘄𝗼𝗿𝗸𝗳𝗹𝗼𝘄𝘀-𝗮𝗰𝗽, a new python package (and CLI toolkit) to build versatile generalist AI agents🚀 The stack: @llamaindex.bsky.social Agent Workflows, with their unmatched power in building powerful event-driven systems🦙

If you're working with Anthropic in @llamaindex.bsky.social, you might want to upgrade to v0.10.5: we just shipped some enhancements in the way we handle structured outputs, and it is now generally more stable and reliable than before (at least in my experience)!🦙

I want to wish you all the Happiest New Year with newyear.clelia.dev and `uv tool install clelia-ny2026`!🥂

New Rust crate alert🦀 I built a small Postgres data viewer in Rust, using sqlx and rich-rs (a rust port for the rich library): securely connect to local and remote Postgres instances, execute your queries and visualize the results as a beautiful table (also with time stats!)🐘

👀 Cooking something with @golang.org, ACP (by @zed.dev) and toad (fantastic work @willmcgugan.bsky.social btw!🙌)

I've been exploring Rust recently, and I built a small web app with Axum (by @tokio.rs), @tur.so Cloud and some HTML and vanilla JS: personalwiki.com.de 🦀

At @llamaindex.bsky.social we recently shipped LlamaSheets in public beta, an agentic and OCR-driven system to carry out advanced processing of spreadsheets🧮

You've just been hired at the North Pole and you already have to save Christmas from a corrupt database!🎄 Luckily, using @llamaindex.bsky.org services such as LlamaClassify, LlamaExtract and Agent Workflows, saving Christmas could not be easier!🎅🦙

I created a CLI tool and MCP server to 𝗴𝗿𝗲𝗽 𝗮𝗻 𝗲𝗻𝘁𝗶𝗿𝗲 𝗱𝗶𝗿𝗲𝗰𝘁𝗼𝗿𝘆 𝗶𝗻 𝗼𝗻𝗲 𝗴𝗼: 𝘥𝘪𝘳𝘨𝘳𝘦𝘱🦫 Built with @golang.org, it concurrently processes all files in a folder to find matches and rich-renders them to terminal🖥️

If your coding agent has ever accidentally deleted important files, you know how hard it is to keep filesystem access under control for AI assistants 🫣

Ever opened a repo and thought: “What does this codebase actually do?” “Where did I put that file?” 🤔 You’re not alone. With the release of Gemini 3 Flash ⚡ from Google DeepMind, we decided to build something fun (and useful): a file-system explorer agent that answers those questions for you. 🔍

Here's the thing, Marjorie: god's not real.

Gemini 3 Fast⚡ is here (the GoogleDeepMind team is really cooking recently) and at @llamaindex.bsky.social we created a fun (and useful) demo showcasing the model's agentic capabilities and speed combined with our Agent Workflows👇

Coding Agents are great, but giving them unfiltered access to your filesystem might be a very bad idea😱

I published my first Rust crate! After following Francesco Ciulla's Rust bootcamp, I decided to get my hands dirty with Rust and create 𝘄𝗼𝗿𝗸𝘀𝗽𝘆𝗰𝗲, a version manager for uv workspaces, inspired by changesets🦋 The project is obviously early stage, but I would love to hear your feedback✨

Day 4 of sharing what's new in PdfItDown V2, the framework for converting (almost) everything to PDF📁 Today is the turn of 🚀deploying PdfItDown as a backend🚀

Day 3 of sharing what's new in PdfItDown V2, the framework for converting (almost) every file to PDF, and today we talk about 👩‍💻CLI👩‍💻 PdfItDown already had a very basic command line interface, and now this evolved into: - Fully-typed options - Rendering with Rich - Faster - Multiple options handling

I just donated to the Python Software Foundation! you should do it too! donate.python.org

This year I decided to take part into #AdventOfCode for the first time, using #golang to solve the challenges. I'm doing this as a personal challenge, so most of my solutions will probably be either brute force or very hacky, but, if you want to follow along 👉 github.com/AstraBert/A...

Day 2 of going through PdfItDown new features, and today we have 💃𝗰𝘂𝘀𝘁𝗼𝗺 𝗰𝗼𝗻𝘃𝗲𝗿𝘀𝗶𝗼𝗻 𝗹𝗼𝗴𝗶𝗰💃

Announcing PdfItDown v2 to the world today! (actually, we are already at 2.0.1!😮) I'll be publishing a series of post about the changes introduced with v2, starting with today's ✨𝗰𝗼𝗱𝗲 𝗾𝘂𝗮𝗹𝗶𝘁𝘆 𝗶𝗺𝗽𝗿𝗼𝘃𝗲𝗺𝗲𝗻𝘁𝘀✨ PS: GitHub repo 👉 github.com/AstraBert/P...

Today is the International Day for the Elimination of the Violence against Women. No better words than the ones used by the song "Figlia d' 'a Tempesta" to describe the struggle and the fight that women all over the world endure every day: youtu.be/486uBfNPkQw 💜✨

PdfItDown (github.com/AstraBert/P...), the python library I created to convert (almost) everything to PDF, is almost at 200 starts⭐ and surpassed 17K all-time downloads📦.

At @llamaindex.bsky.social we created StudyLlama, a small demo app to show how you can get LlamaAgents to organize your notes📝

🎥 Take a look at the demo, where @cle-does-things.bsky.social explains you all these features and more: www.loom.com/share/c12d4... 👩‍💻 Check our the GitHub repo: github.com/run-llama/s... 🦙 Get started with LlamaCloud: https://cloud.llamaindex.ai?utm_source=socials&utm_medium=li_social

📊 Create categories to classify your notes 📓 Upload your notes 🤖 Watch as LlamaClassify assigns one of the categories to them and LlamaExtract produces a summary and a set of questions-and-answers 🗄️ The extracted info is uploaded to a vector database 🔍 Ask questions to your notes and get answers

Have a LlamaAgent organize your study material: meet StudyLlama, a web app that uses LlamaAgents to help you organize and gather insights from your notes and papers! How it works:

As we remember all our brothers and sisters that were forced out of existence today, my heart goes out to everyone who's fighting to stay. Remember that for everyone who hates you, there will always be many more who love you and root for you. Keep fighting, keep shouting, keep existing🏳️‍⚧️

As developers, there is no brag that compares to a very green @github.com contributions graph. Well, I think every developer should celebrate their GitHub Year, no matter how green their little squares are: that's why I built gityear.re! 🎉

Observability in AI is essential, especially for agentic apps handling unstructured documents. Without it, your workflows become fragile black boxes. At @llamaindex.bsky.social, instrumentation is built in, ready to plug into your existing tracing pipeline.

LlamaAgents is now in open preview - the fastest way to build, serve, and deploy multi-step document agents that combine LlamaCloud's document extraction and parsing power with Agent Workflows orchestration.

The Google DeepMind team just dropped Gemini 3, and we at LlamaIndex have day-zero support! We also made a little demo to show how you can leverage the advanced agentic capabilities of Gemini 3 to automate your GitHub workflow around PRs, you just need to run: 𝘱𝘪𝘱 𝘪𝘯𝘴𝘵𝘢𝘭𝘭 𝘱𝘳-𝘮𝘢𝘯𝘢𝘨𝘦𝘳

Yesterday we announced the open preview for LlamaAgents So, me and my colleague Adrian wrote this intro blog to help you get started. Learn about all the document agent templates available to you via llamactl and:

The team at Google DeepMind cooked🔥 Gemini 3 is honestly one of the best models I've seen so far, and at @llamaindex.bsky.social we offer day-zero support for it.

I've been exploring observability and monitoring a lot recently, and I've been particularly loving @posthog.com. That's why I came up with a way to pipe @opentelemetry.io events directly to it👇 𝘱𝘪𝘱 𝘪𝘯𝘴𝘵𝘢𝘭𝘭 𝘱𝘰𝘴𝘵𝘩𝘰𝘨-𝘰𝘵𝘦𝘭-𝘦𝘹𝘱𝘰𝘳𝘵𝘦𝘳 GitHub: github.com/AstraBert/p...

🎉 What a Git Year I had! I created 34 public repos, gained 144 stars and 17 forks and pushed 1176 public commits! ➡️ Follow me on GitHub: github.com/Goodnight77 Tried out @cle-does-things.bsky.social’s GitYear app and it nailed it Check it out if you want your own GitHub recap gityear.re

Chart OCR just got a major upgrade with our new experimental "agentic chart parsing" feature in LlamaParse 📈🧪

Trigger your agent workflows directly from your inbox, using our LlamaAgents and @resend.com webhooks📧 In this demo, we built a system that:

Here's a common scenario: Your finance team gets emails all day with invoices from partners and expense reports from employees. Each one needs different handling. Invoices need acknowledgment and payment scheduling. Expenses need budget validation before approval etc.

If you wanna hear me yap for 15 minutes about the two things that excite me the most in the AI space right now, i.e. @llamaindex.bsky.social agent workflows and vector search, and how together they can create context-rich AI systems, here's my talk at @qdrant.bsky.social VSD: youtu.be/CDyFukgpayY

Augment your LlamaIndex agent workflows with memory and persistent states: Check out @cle-does-things.bsky.social's talk at @qdrant.bsky.social Vector Space Day to learn how to build context-rich AI apps leveraging vector search and workflow engineering.

Do you want to hear about my first time (with Cluade Code)? Well, you're in luck, since at @llamaindex.bsky.social we just recorded a tutorial of how to vibe-code your way to production with our agent workflows, using Claude code.

Our OSS engineer @cle-does-things.bsky.social tried using Claude Code with high-quality documentation and skills provided by vibe-llama (our tool for the vibe-coding ecosystem), and in this video shows you how to create an agent workflow for financial document classification and deploy to the cloud

Hey, I came up with a small PR, I think we can merge it without review...