iiinova — where your digital self keeps evolving

Clelia Astra Bertelli @cle-does-things.bsky.social · 05/07/2026

PdfItDown has WASM bindings that provide you with fast conversion to PDF in the browser and on the edge. You can get started by deploying a @cloudflare.social worker with less than 20 lines of code, using the snippet below👇 Full example: github.com/AstraBert/P...

000

Clelia Astra Bertelli @cle-does-things.bsky.social · 17/06/2026

Write your first agent in Rust with 𝗺𝗶𝗰𝗿𝗼𝗮𝗴𝗲𝗻𝘁𝘀🦀 Define a provider, create custom tools, load skills from local or global .𝚊𝚐𝚎𝚗𝚝𝚜/𝚜𝚔𝚒𝚕𝚕𝚜, add instructions, and stream agent events in real time🤖 Get started: microag.clelia.dev ⭐ the GitHub repo: github.com/AstraBert/m...

010

Clelia Astra Bertelli @cle-does-things.bsky.social · 01/06/2026

Happy Pride Month!!!🏳️‍🌈🏳️‍⚧️ I hope each and every one of you will be able to feel how loved you are every day, by me, by the community and by all the allies out there💖

030

Clelia Astra Bertelli @cle-does-things.bsky.social · 24/05/2026

Install the latest version with 𝘯𝘱𝘮 𝘪 -𝘨 @𝘤𝘭𝘦-𝘥𝘰𝘦𝘴-𝘵𝘩𝘪𝘯𝘨𝘴/𝘧𝘶𝘴𝘦𝘱𝘳𝘰𝘹𝘺, connect it to your Langfuse setup (I have a locally self-hosted instance), and enjoy in-depth tracing without hassles :) ⭐ GitHub repo: github.com/AstraBert/f...

github.com

GitHub - AstraBert/fuseproxy: API proxy for coding agents with Langfuse-powered instrumentation

API proxy for coding agents with Langfuse-powered instrumentation - AstraBert/fuseproxy

010

Clelia Astra Bertelli @cle-does-things.bsky.social · 24/05/2026

Of course, I want to track costs, usage and input/outputs, so I adapted fuseproxy, the Langfuse-instrumented proxy I built for th Anthropic Messages API, to natively integrate with Pi.

100

Clelia Astra Bertelli @cle-does-things.bsky.social · 24/05/2026

My latest setup for agentic engineering in personal projects includes Copilot and Pi powered by Kimi K2.6.

100

Reposted by Clelia Astra Bertelli

Andri Óskarsson @andri.dk · 20/05/2026

I miss programming by hand. My day-job is now project-leading something that acts like a know-it-all junior developer on cocaine. Worst part is, that if you look past the obvious security issues it's the most productive I've ever been.

5767748

Reposted by Clelia Astra Bertelli

Albert Pinto @70sbachchan.bsky.social · 17/05/2026

daaaamn london

116176644692

Clelia Astra Bertelli @cle-does-things.bsky.social · 15/05/2026

I wrote about both of these approaches to coding agent observability in my latest blog post: clelia.dev/2026-05-14-... Hope you enjoy the read, and I’d love to hear your thoughts👇

clelia.dev

Instrumenting Coding Agents

Yapping about adding an observability layer to coding agents

020

Clelia Astra Bertelli @cle-does-things.bsky.social · 15/05/2026

I also built instrument-claude, an observability tool that turns Claude Code's hook events into OTel-compatible traces, spans, and events, providing an overview of everything happening within a conversation turn or across an entire session.

100

Clelia Astra Bertelli @cle-does-things.bsky.social · 15/05/2026

I built fuseproxy, a @langfuse.bsky.social -instrumented API proxy that coding agents can call instead of talking directly to provider servers, giving me visibility into usage, costs, and everything flowing in and out of the agent.

210

Clelia Astra Bertelli @cle-does-things.bsky.social · 15/05/2026

Coding agents might feel like magic, but they aren’t 🪄 That’s exactly why I’ve been eager lately to understand how they work under the hood, using the most natural approach I could think of: observability.

130

Clelia Astra Bertelli @cle-does-things.bsky.social · 14/05/2026

- a much better understanding of how the agent behaves internally Huge difference if you don’t want your coding agents operating as black boxes. Code here: github.com/AstraBert/f...

github.com

GitHub - AstraBert/fuseproxy: API proxy for coding agents with Langfuse-powered instrumentation

API proxy for coding agents with Langfuse-powered instrumentation - AstraBert/fuseproxy

000

Clelia Astra Bertelli @cle-does-things.bsky.social · 14/05/2026

Best part? Everything runs locally and fully self-hosted with Docker Compose. This setup gives me: - visibility into what the agent actually sends and receives - insight into tool calls and tool outputs - token + cost tracking - latency monitoring

100

Clelia Astra Bertelli @cle-does-things.bsky.social · 14/05/2026

- Express backend running on Bun - Anthropic SDK integration with full streaming support - Plug-and-play observability via LangFuse for tracing inputs, outputs, tokens, latency, and costs - Integrated with GitHub Copilot (my favorite coding agent right now)

100

Clelia Astra Bertelli @cle-does-things.bsky.social · 14/05/2026

Coding agents shouldn’t be black boxes ❌ If you’re calling models through APIs and can configure provider settings (base URL, API key, etc.), I highly recommend putting a local instrumented API proxy in front of them. I built one for the Anthropic Messages API:

230

Clelia Astra Bertelli @cle-does-things.bsky.social · 12/05/2026

We also included ready-to-use examples for rate limiting, caching, and OpenTelemetry-compatible traces and metrics collection with tools like Redis, Jaeger, Prometheus and Grafana🔭

100

Clelia Astra Bertelli @cle-does-things.bsky.social · 12/05/2026

It’s 100% open source, fully self-hostable, and your data stays yours✅ Built in TypeScript with ExpressJS, liteparse-server can run as a Docker container or in serverless environments.

100

Clelia Astra Bertelli @cle-does-things.bsky.social · 12/05/2026

Do you actually own your document parsing infrastructure? 👀 At LlamaIndex, we wanted to make that easier, so we built 𝗹𝗶𝘁𝗲𝗽𝗮𝗿𝘀𝗲-𝘀𝗲𝗿𝘃𝗲𝗿, a lightweight HTTP backend built on top of LiteParse that can parse and generate page screenshots from PDFs, images, and Office documents🦙

110

Clelia Astra Bertelli @cle-does-things.bsky.social · 11/05/2026

Once the results are ready, I can cleanly push the final changes to GitHub. Learn more → aggit.dev Install as an agent skill → 𝘯𝘱𝘹 𝘴𝘬𝘪𝘭𝘭𝘴 𝘢𝘥𝘥 𝘈𝘴𝘵𝘳𝘢𝘉𝘦𝘳𝘵/𝘢𝘨𝘨𝘪𝘵 Leave a ⭐ → github.com/AstraBert/a...

github.com

GitHub - AstraBert/aggit: S3-backed, git-versioned object storage for your agents

S3-backed, git-versioned object storage for your agents - AstraBert/aggit

000

Clelia Astra Bertelli @cle-does-things.bsky.social · 11/05/2026

With aggit, my agents can stash intermediate work, create branches safely, restore previous states, and back up artifacts to my RustFS server with multi-instance replication support, all without touching the main Git repository.

110

Clelia Astra Bertelli @cle-does-things.bsky.social · 11/05/2026

So I built 𝗮𝗴𝗴𝗶𝘁: a Git-like CLI for local and remote (S3-backed) agent artifact storage, written in Rust 🦀.

100

Clelia Astra Bertelli @cle-does-things.bsky.social · 11/05/2026

I have one big problem with agentic engineering: I want agents to operate autonomously, but I also want granular, reversible control over every change they make. I could solve this by committing every intermediate step to Git, but that would completely pollute my repo history.

110

Clelia Astra Bertelli @cle-does-things.bsky.social · 01/05/2026

PS: I'm thinking about writing a blog about this, would someone be interested in reading it?👀

010

Clelia Astra Bertelli @cle-does-things.bsky.social · 01/05/2026

🌐 Axum REST APIs to interact with Redis and Qdrant ➕ Observability with Jager and @opentelemetry.io, rate limiting with Memcached 👩‍💻 And, as always, the code is open source if you want to customize it: github.com/AstraBert/f...

github.com

GitHub - AstraBert/file-storage: Self-hostable file storage platform built on a modern open-source stack

Self-hostable file storage platform built on a modern open-source stack - AstraBert/file-storage

100

Clelia Astra Bertelli @cle-does-things.bsky.social · 01/05/2026

🐳 Simple @docker.com setup using compose 🔒 Authentication with Keycloak 🗄️ S3 storage with Garage, exposed through a Tonic GRPC server ⚡ File metadata storage in @redis.io 🔍 A background worker that indexes your file metadata for search, powered by @rabbitmq.com and @qdrant.bsky.social

110

Clelia Astra Bertelli @cle-does-things.bsky.social · 01/05/2026

For 2026, I decided I'm going to build all of my side projects with Rust when possible🦀 So here's the first one: 𝗳𝗶𝗹𝗲-𝘀𝘁𝗼𝗿𝗮𝗴𝗲, a self-hostable file storage application built on a modern open source stack, with a full Rust backend and a @react.dev + @vite.dev frontend🚀 With file-storage, you get:

130

Clelia Astra Bertelli @cle-does-things.bsky.social · 30/04/2026

The interesting part here is not the single step, but how we orchestrate all the moving parts without introducing bottlenecks. Read the full breakdown 👉 render.com/blog/buildi...

render.com

Building Document Pipelines That Actually Scale

Create a distributed processing engine with LlamaIndex and Render Workflows.

100

Clelia Astra Bertelli @cle-does-things.bsky.social · 30/04/2026

* parsing, classification, and extraction with @llamaindex.bsky.social (via LlamaParse) 🦙 * distributed task execution using Render Workflows♻️ * a lightweight server + database layer to interact with the pipeline in real time⚡

100

Clelia Astra Bertelli @cle-does-things.bsky.social · 30/04/2026

Scaling document pipelines is harder than it looks, and not because of tooling, but mostly because of orchestration⚙️ That's why I teamed up with @render.com and focused on building a system where each component does one job well, and scales independently:

100

Clelia Astra Bertelli @cle-does-things.bsky.social · 29/04/2026

📝 Of course, building a production MCP server means encountering challenges along the way, and you can read about them all in the blog post we wrote: www.llamaindex.ai/blog/llamap... 👩‍💻 GitHub repo: github.com/run-llama/m...

github.com

GitHub - run-llama/mcp-llamaindex-ai: Our own point-and-click MCP server.

Our own point-and-click MCP server. Contribute to run-llama/mcp-llamaindex-ai development by creating an account on GitHub.

000

Clelia Astra Bertelli @cle-does-things.bsky.social · 29/04/2026

We refactored our MCP to have: - Direct integration with our Parse, Classify and Split services🦙 - A smoother authentication flow using WorkOS🔒 - Seamless support for file uploads⬆️ - Observability, rate-limiting and fast deployments with @vercel.com and @axiom.co

100

Clelia Astra Bertelli @cle-does-things.bsky.social · 29/04/2026

The LlamaParse MCP got a new face, and it is now easier than ever to run document processing workflows from your agents🚀

100

Clelia Astra Bertelli @cle-does-things.bsky.social · 21/04/2026

Thanks to everyone who helped in getting me on board with this! PS: link for tickets in the next comment👇

100

Clelia Astra Bertelli @cle-does-things.bsky.social · 21/04/2026

I'll be speaking at the Applied AI Conf organized by Tech Europe!🇪🇺 On the 28th of May, I'll be in Berlin to give my talk about how you can build reliable document agents by leveraging harness engineering with the building blocks provided by @llamaindex.bsy.social 🦙

100

Clelia Astra Bertelli @cle-does-things.bsky.social · 18/04/2026

I'll be speaking at the upcoming Voxel51 event in Stuttgart this Tuesday!🚀 I'll talk about the anatomy of AI agents, with a focus on document agents and building good harnesses for them⚙️ Swing by if you're interested, and check the official page for more details: voxel51.com/events/stut...

voxel51.com

Stuttgart AI, ML and Computer Vision Meetup - April 21, 2026

Join our in-person meetup to hear talks from experts on cutting-edge topics across AI, ML, and computer vision.

020

Clelia Astra Bertelli @cle-does-things.bsky.social · 17/04/2026

📝I decided to write a blog post about it, mostly to document the building journey for myself, but also to share my thinking and coding process around sunbears: clelia.dev/blog/2026-0... Enjoy!✨

100

Clelia Astra Bertelli @cle-does-things.bsky.social · 17/04/2026

I've been building sunbears, a typescript library for CSV data loading written in Rust🦀

110

Clelia Astra Bertelli @cle-does-things.bsky.social · 13/04/2026

What makes it different? ParseBench optimizes for semantic correctness, not exact text matching. That means evaluating whether parsed outputs are actually useful for humans and AI agents making downstream decisions 🔍

210

Clelia Astra Bertelli @cle-does-things.bsky.social · 13/04/2026

It includes: • 2,000+ human-reviewed enterprise documents • 167,000 evaluation rules • Coverage across 5 key areas: tables, charts, content faithfulness, semantic formatting, and visual grounding

100

Clelia Astra Bertelli @cle-does-things.bsky.social · 13/04/2026

ParseBench is here!📊 We’ve just released ParseBench, an open benchmark + dataset for evaluating document parsing at scale.

100

Clelia Astra Bertelli @cle-does-things.bsky.social · 08/04/2026

sunbears uses 𝘋𝘢𝘵𝘢𝘍𝘳𝘢𝘮𝘦 as its primary data structure, a columnar format with strict typing, null and NaN filtering, and convenient column-to-array transformations📄 📦 Get started now: 𝘯𝘱𝘮 𝘪𝘯𝘴𝘵𝘢𝘭𝘭 @𝘤𝘭𝘦-𝘥𝘰𝘦𝘴-𝘵𝘩𝘪𝘯𝘨𝘴/𝘴𝘶𝘯𝘣𝘦𝘢𝘳𝘴

100

Clelia Astra Bertelli @cle-does-things.bsky.social · 08/04/2026

I'm building 𝘀𝘂𝗻𝗯𝗲𝗮𝗿𝘀, a CSV data loader library for TS written in Rust🦀 In Node, it can read a file with 1.000.000 rows in 0.3s, write the same amount of rows in 0.15s, respectively 4x and 2x faster than the `csv` package⚡

110

Reposted by Clelia Astra Bertelli

LlamaIndex @llamaindex.bsky.social · 07/04/2026

Visually rich documents are especially challenging for agents. Tables, charts, and images often break traditional document pipelines, making complex reasoning difficult📄 So we teamed up with LanceDB to build a structure-aware PDF QA pipeline🚀 Here’s how it works:

211