A fast and local neural text-to-speech engine that embeds espeak-ng for phonemization.

AGENTS.md

A simple, open format for guiding coding agents,
used by over 60k open-source projects.

Think of AGENTS.md as a README for agents: a dedicated, predictable place to provide the context and instructions to help AI coding agents work on your project.

An Eye on your system

Glances is a cross-platform system monitoring tool written in Python

Saviez-vous que la pagination par offset est très complexe mais facile à éviter ?

La clause offset demande à la base de données d'ignorer les N premiers résultats d'une requête. Cependant, la base de données doit toujours récupérer ces lignes du disque avant de pouvoir renvoyer les lignes suivantes.

Ce n'est pas un problème provenant de l'implémentation, c'est à la base du concept même d'offset :

… les lignes sont tout d'abord triées suivant la <clause order by>, puis sont filtrées en supprimant le nombre de lignes indiqué dans la <clause offset> à partir du début…

SQL:2023, Part 2, §4.17.3 Derived tables

Autrement dit, les gros décalages imposent un long travail à la base de données, qu'elle soit SQL ou NoSQL.

Text-to-speech

Kyutai text-to-speech started as an internal tool we used during the development of Moshi. As part of our commitment to open science, we've since open-sourced two text-to-speech models:

Kyutai Pocket TTS, a tiny model with voice cloning, fast enough to run on CPU.
Kyutai TTS 1.6B, a streaming model used in Unmute, great for servers.

Popeye: Kubernetes Live Cluster Linter

Popeye is a utility that scans live Kubernetes clusters and reports potential issues with deployed resources and configurations. As Kubernetes landscapes grows, it is becoming a challenge for a human to track the slew of manifests and policies that orchestrate a cluster. Popeye scans your cluster based on what's deployed and not what's sitting on disk. By linting your cluster, it detects misconfigurations, stale resources and assists you to ensure that best practices are in place, thus preventing future headaches. It aims at reducing the cognitive overload one faces when operating a Kubernetes cluster in the wild. Furthermore, if your cluster employs a metric-server, it reports potential resources over/under allocations and attempts to warn you should your cluster run out of capacity.

Popeye is a readonly tool, it does not alter any of your Kubernetes resources in any way!

OpenCensus and OpenTracing have merged to form OpenTelemetry, which serves as the next major version of OpenCensus and OpenTracing.

OpenTelemetry has now reached feature parity with OpenCensus, with tracing and metrics SDKs available in .NET, Golang, Java, NodeJS, and Python

Optimisation webperf : AVIF et pré-compression pour le blog

The High-Throughput and Memory-Efficient inference and serving engine for LLMs

mise-en-place

The front-end to your dev env

to replace asdf ?

Expensively Quadratic: the LLM Agent Cost Curve
2026-02-03 Philip Zeyliger

Pop quiz: at what point in the context length of a coding agent are cached reads costing you half of the next API call? By 50,000 tokens, your conversation’s costs are probably being dominated by cache reads.

Let’s take a step back. We’ve previously written about how coding agents work: they post the conversation thus far to the LLM, and continue doing that in a loop as long as the LLM is requesting tool calls. When there are no more tools to run, the loop waits for user input, and the whole cycle starts over.