Would you elaborate?

Ships with ready-to-use example skill packages:
- rag — search & analyze documents via haiku.rag
- web — search via Brave Search API + page extraction via trafilatura
- code-execution — sandboxed Python via pydantic-monty
- image-generation — generate images via Ollama

Overview - Agent Skills A simple, open format for giving agents new capabilities and expertise.

Features:
- Discover skills from SKILL.md directories or Python entrypoints
- Wrap any MCP server (stdio, SSE, streamable HTTP) as a skill
- Attach in-process Python tools or full toolsets
- Script tools with inline dependencies via PEP 723
- Progressive disclosure

haiku.skills — give your AI agent skills without drowning it in tools. Each skill runs in its own sub-agent with dedicated
instructions, tools, and token budget.

Built on @pydantic.dev's pydantic-ai, follows the Agent Skills spec. github.com/ggozad/haiku.skills

Text classification with Python 3.14's zstd module Python 3.14 introduced the compression.zstd module. It is a standard library implementation of Facebook’s Zstandard (Zstd) compression algorithm. It was developed a decade ago by Yann Collet, who hold...

maxhalford.github.io/blog/text-cl...

This is a cool blog post about using compression.zstd in Python as a surprisingly effective (and fast!) simple text classifier.

GitHub - ggozad/haiku.rag: Retrieval Augmented Generation based on LanceDB Retrieval Augmented Generation based on LanceDB. Contribute to ggozad/haiku.rag development by creating an account on GitHub.

haiku.rag 0.16.0 is here! 🎉

Build research agents with full AG-UI protocol support:
Stream graph execution with real-time progress
Track questions, answers & insights as they unfold
Full-stack example with @pydantic.dev AI + CopilotKit

Turn RAG into a research engine.

github.com/ggozad/haiku...

GitHub - ggozad/haiku.rag: Retrieval Augmented Generation based on LanceDB Retrieval Augmented Generation based on LanceDB. Contribute to ggozad/haiku.rag development by creating an account on GitHub.

I made an example app using my haiku.rag (github.com/ggozad/haiku...), the open-source local RAG library using @pydantic.dev, to create a research assistant.
Video demonstration here: vimeo.com/1128874386

#AI #python #open-source #RAG

ΟΜΗΡΙΑΣ & ΕΓΚΛΗΜΑΤΟΣ ΣΥΝΕΧΕΙΑ! Το Ισραήλ αρνείται να απελευθερώσει τον Δρ. Hussam Abu Safiya, ο οποίος επέμεινε να μείνει στο πλευρό των ασθενών του στο βομβαρδισμένο νοσοκομείο του, ακόμη και μετά τη δολοφονία του ίδιου του γιου του. Από τότε κρατείται σε ένα από τα κέντρα βασανιστηρίων του Ισραήλ.

....ορθή επανάληψη το βιντεάκι που ανέβασα και πριν για να το βλέπετε και χωρίς λογαριασμό στο ινστα....
...όταν οι παράδοσεις ,ιδίως όσες αφορούν γυναίκες, είναι απλά πατριαρχία του χειρότερου είδους.....

haiku.rag, is a RAG library that can use open-source local models or commercial providers without the need for external vector databases.

From 0.7.0 we moved to be using LanceDB (lancedb.com), gaining performance and reaching 85% accuracy (from 67%) on basic settings.

github.com/ggozad/haiku...

The Freedom Flotilla announces its intention to set sail again:

This time dozens of boats;
Coordinated from 44 countries;
Carrying aid, food, medicine & doctors;

The 18 year siege of Gaza 🇵🇸 to be broken:

Opposing Israel
Freeing Palestine 🇵🇸
Supporting humanity.

#FreePalestine

GitHub - ggozad/haiku.rag: Retrieval Augmented Generation based on SQLite Retrieval Augmented Generation based on SQLite. Contribute to ggozad/haiku.rag development by creating an account on GitHub.

Introducing haiku.rag - A SQLite-based RAG client & server that keeps your data local!

- Pure SQLite - no servers needed
- Hybrid search
- Multiple embedding providers (Ollama, OpenAI, VoyageAI)
- Auto file monitoring
- MCP server

github.com/ggozad/haiku...

#RAG #SQLite #AI #OpenSource #MCP

New in this release:
- In addition to oterm you can use oterm-command to create your own custom CLI commands based on chats.
- oterm now fully supports Model Context Protocol (MCP) tools.

#ollama #python #llm #tui

GitHub - ggozad/oterm: a text-based terminal client for Ollama a text-based terminal client for Ollama. Contribute to ggozad/oterm development by creating an account on GitHub.

I am happy to announce oterm 0.9.1!
oterm is a featureful terminal client for Ollama, where you can easily customize and chat with any Ollama model & tools.

Github: github.com/ggozad/oterm
Docs: ggozad.github.io/oterm/

GitHub - ggozad/oterm: a text-based terminal client for Ollama a text-based terminal client for Ollama. Contribute to ggozad/oterm development by creating an account on GitHub.

Happy new year with a new release for oterm, the best open-source terminal client for #Ollama.

0.7.2 is a major update, including:
* Structured output support.
* Ability to add user/custom tools for LLMs that support it.

Looking forward to what 2025 will bring 🎉
github.com/ggozad/oterm
#LLM #AI

oterm - A text-based terminal client for ollama. A text-based terminal client for ollama. - Discover more CLI and TUI tools like oterm and much more on Terminal Trove.

oterm is a TUI client for Ollama.

It supports multiple chat sessions, copying messages & keyboard shortcuts. Any LLMs from ollama's library can run in oterm's TUI.

oterm is made by @ggozad.bsky.social with the Textual framework & is Terminal Tool of the Week! ⭐️

terminaltrove.com/oterm/

A picture of a llama shown inside oterm.

Announcing oterm 0.1.17 release. oterm is a terminal UI for ollama. Latest version bring support for multi-modal models, so you can now use images in your chats, without ever leaving your terminal! Grab it here: github.com/ggozad/oterm
#llm #ml #ollama #python #textual

I wrote oterm a text-based terminal app that allows you to chat with your Ollama models. It’s fast, easy, remembers your conversations, can use any Ollama model custom or not. Grab it at github.com/ggozad/oterm #python #llm #ollama