Linux Blimp

Meet dana — a free, open-source data analysis terminal for Linux.

root — Tue, 09 Jun 2026 03:57:08 +0000

Meet dana — a free, open-source data analysis terminal for Linux.

Type a command. Get an interactive chart. Let AI explain what it means.

What dana can do:

• Stocks — price + volume charts, up to 10 years of history
• Crypto — Bitcoin, Ethereum, Solana and more
• Currency — exchange rates for any pair
• Weather — temperature, precipitation, wind (Metric or Imperial)
• Air Quality — PM2.5, NO₂, Ozone — hourly data
• GDP — World Bank data, up to 60 years
• Live Flights — real-time aircraft positions on a map
• Weather Maps — city temperatures across any region

All of this with multi-series comparison — put AAPL, TSLA and NVDA on one normalized chart, compare London vs Paris vs Moscow weather, track USD against EUR, GBP and JPY simultaneously.

AI Analysis — the part I’m most proud of:

After every fetch, dana automatically sends the data to your AI model and displays the analysis right below the chart. Trends, anomalies, comparisons, one-sentence conclusion.

Works in 15 languages: English, Russian, Chinese, Japanese, Spanish, French, German, Italian, Portuguese, Arabic, Korean, Hindi, Turkish, Polish, Dutch. Light and Dark themes. 3 any of mshell compatible LLMs by your choice (Claude, chatGPT, Gemini, any ollama models).

Your data. Your model. Your language.

Built with C + GTK4 + Python + Plotly. All data sources are free and open — Open-Meteo, CoinGecko, OpenSky, World Bank, yfinance.

MIT License. One install script. Works on Ubuntu 22.04/24.04/26.04 and Debian 13.

github.com/igor101964/dana

If you find it useful — star it, share it, or open an issue. Contributions welcome.

#opensource #linux #dataanalysis #python #gtk #finance #ai #freetools

LLMs understand humor.

root — Tue, 09 Jun 2026 03:53:38 +0000

Turns out LLMs can be funny.

I gave three different AI models the same prompts — draw a developer at 3am, a cat on a Roomba, Einstein arguing with Newton — and collected the results.

No model names. No rankings. Just ASCII art and whatever passes for a sense of humor when you have no eyes and were trained on the internet.

Some results are surprisingly good. Some are gloriously bad. All of them tried.

“Models with Humor… as best they can” — a small ASCII comic booklet generated with nomo, our open-source GTK4 ASCII art tool.

nomo on GitHub: https://lnkd.in/g75hgZWS

#ASCII #LLM #OpenSource #Art2DecSoftLab #nomo #humor

Meet NOMO – Open-source ASCII Art Generator in C/GTK4 — LLM-powered + local Figlet rendering.

root — Tue, 09 Jun 2026 03:48:53 +0000

Meet NOMO – Open-source ASCII Art Generator in C/GTK4 — LLM-powered + local Figlet rendering!

Built from scratch as part of the Art2Dec SoftLab open-source initiative. Runs natively on Linux x86_64 and ARM64 (Raspberry Pi 4/5).

Two modes:
• **Figlet mode** — instant local text rendering using any installed font, no internet needed
• **LLM mode** — generate ASCII art and comic strips via mshell IPC (Claude, ChatGPT, Gemini, Ollama)

3 configurable LLM model slots — read directly from ~/.mshellrc
10 included system prompts — art, banners, comics with speech bubbles
Comic strip generator — characters with funny dialogue invented by the model
Live token streaming — output appears as the model generates
Dark / Light theme, art color selector (8 colors)
Inline system prompt editor — tweak prompts without leaving the app
Copy to clipboard — works natively in GTK4
Multi-language input — English, Russian and more

Tech stack:
**C / GTK4** — native Linux desktop UI
**Figlet** — local ASCII font rendering
**mshell** — LLM workflow platform by Art2Dec SoftLab

Tested platforms:
• Ubuntu 22.04 / 24.04 / 26.04 — x86_64
• Debian 12 / 13 — x86_64, ARM64 (Raspberry Pi 4b)

MIT License — free to use, modify, and distribute.

GitHub: https://lnkd.in/g75hgZWS

#OpenSource #C #GTK4 #Linux #RaspberryPi #ASCII #LLM #Art2DecSoftLab #mshell #Figlet #DevTools

MIDE is now running on Ubuntu 26.04 LTS (Resolute Raccoon).

root — Tue, 09 Jun 2026 03:45:28 +0000

MIDE is now running on Ubuntu 26.04 LTS (Resolute Raccoon)! MIDE — the primary development workstation of the mshell Ecosystem — has been successfully tested and deployed on the latest Ubuntu LTS, Ubuntu 26.04 LTS “Resolute Raccoon”. The entire mshell Ecosystem stack (mshell, mide, mel, mshell Workflow, Studio, edi, Notebook) now officially runs on the current platform and all current Ubuntu derivatives and continue to support Ubuntu 22.04-24.04 stack as well like Debian12/13 on different platforms including ARM64.

Keyboard Shortcuts Cheatsheet for MIDE — the GTK4 development workstation for the mshell Ecosystem.

root — Tue, 09 Jun 2026 03:40:40 +0000

Keyboard Shortcuts Cheatsheet for MIDE — the GTK4 development workstation for the mshell Ecosystem.

**MIDE** is a polyglot IDE written in C with native AI integration — multi-tab editor, embedded mshell & bash terminals, LSP support, LLM Explain/Fix/Generate, AI Chat, image generation, and full mshell Workflow support. Built for Linux x86_64.

The cheatsheet covers all key areas:

File & Project
Editor
Search
AI
Terminals (mshell & bash)
Other

Art2Dec SoftLab · mshell Ecosystem
#mide #mshell #Art2DecSoftLab #IDE #LLM #AI #GTK4 #Linux #OpenSource #DevTools

Keyboard Shortcuts Cheatsheet for mel

root — Tue, 09 Jun 2026 03:31:05 +0000

Keyboard Shortcuts Cheatsheet for mel — a lightweight terminal text editor for the mshell Ecosystem.

**mel** is a minimalistic terminal editor written in pure C — no curses, 102KB binary, full UTF-8, 15 languages syntax highlighting, and native LLM integration. Runs on Linux, macOS, and Raspberry Pi.

The cheatsheet covers all key areas:

Keybindings
Search & Navigation
Options
Syntax highlighting (15 languages)

Art2Dec SoftLab · mshell Ecosystem · github.com/igor101964/mel
#mel #mshell #Art2DecSoftLab #TerminalEditor #LLM #Linux #macOS #RaspberryPi #OpenSource #DevTools

MEL – terminal text editor with native LLM integration

root — Wed, 03 Jun 2026 21:00:17 +0000

Meet mel — probably the world’s smallest terminal text editor with native LLM integration.

102KB. Pure C. No curses. No dependencies beyond libcurl.
And yes — it talks to Ollama, ChatGPT, Claude, and Gemini right from the editor buffer.
Built from scratch at Art2Dec SoftLab as part of the mshell Ecosystem.
For larger projects — meet mide, a full GTK4 IDE with the same LLM integration, file tree, multi-tab editor, terminal, and AI chat built in.

What makes mel different:
102KB binary — fits in any container, pod, or IoT device
Also available as a fully static binary (6.1MB, zero dependencies) — drop it anywhere, no install needed
15 languages syntax highlighting — C/C++, Rust, Go, Python, Java, Bash, mshell, JS, JSON, SQL, PHP, XML, Ruby, Lua
Full UTF-8 — Russian, Japanese, Chinese, CJK — correct cursor, backspace, and display
Native LLM integration:
• Ctrl-W — full-screen prompt editor → sends to LLM switch router via mshell, inserts response at cursor
• Ctrl-K — connects to any local or remote Ollama-compatible model via simple config file
• Works with Ollama, ChatGPT, Claude, Gemini — whatever your stack uses
System clipboard — Ctrl-T paste from xclip/xsel/pbpaste
Undo/Redo · Search & Replace · Go to line · Column marker · Backup

Built for everywhere:

Kubernetes Pods and Docker containers — single binary, zero setup
Cloud VMs — AWS, GCP, OCI — x86_64 and ARM64
IoT and embedded — Raspberry Pi 4/5 ARM64 (Debian 12/13)
Desktop — Ubuntu 22.04/24.04 LTS, macOS Sequoia (Intel & Apple Silicon)
Works in any POSIX shell — bash, zsh, sh, mshell

Tech:
Pure C (C99) — single source file, no curses, direct VT100 terminal control
libcurl + libjson-c for LLM REST API integration
GPL3 license — free to use, modify, and distribute

Documentation & binaries on GitHub:
GitHub: https://github.com/igor101964/mel
• Keyboard Shortcuts cheatsheet (PNG/PDF) → /cheatsheets
• Full User Guide PDF → /docs
• Pre-built binaries for all above platforms → /binaries

OpenSource #C #Linux #RaspberryPi #macOS #TerminalEditor #LLM #Ollama #ChatGPT #Claude #Gemini #mshell #Art2DecSoftLab #DevTools #Kubernetes #Docker #IoT #UTF8

mide + mshell Workflow Role.

root — Sat, 30 May 2026 23:07:17 +0000

mide + mshell Workflow Role = Accidental Superpower Here’s a fun side effect of how mide and mshell Workflow roles work together. When you load mshell_expert.role as a system prompt in AI Chat, the LLM is instructed to wrap every piece of code in fenced code blocks — exactly the mshell Workflow pipeline format. So when you ask it to “write Rust code that draws sine and cosine in ASCII”, it does two things at once without being asked: Puts the Rust source into the editor via Generate Puts the same code as a workflow block into LLM Output — ready to run in edi Now you have two independent execution paths from a single prompt: Run button → mide compiles and runs the Rust file → result appears in the Output panel edi button → loads the code from LLM Output → executes it as a mshell Workflow stage → result appears in the mshell VTE terminal Same code. Same sine wave. Two completely different runtimes — both work perfectly. This isn’t a bug. The role does exactly what it’s told. The LLM just happens to speak fluent mshell Workflow by default when the role is loaded — and mide is ready to run whatever comes out, in whatever way you prefer. Moral of the story: write your roles right, and the pipeline takes care of itself.

MIDE — the primary development workstation for the mshell Ecosystem that brings together a polyglot editor, AI assistance, simultaneous LLM models, and the full mshell toolchain.

root — Sat, 30 May 2026 23:01:47 +0000

Meet MIDE — the primary development workstation for the mshell Ecosystem that brings together a polyglot editor, AI assistance, simultaneous LLM models, and the full mshell toolchain.

Editor & Built-in Tools:
Multi-tab GtkSourceView editor — C, C++, Go, Rust, Python, Lua, Bash, mshell
Syntax highlighting, line numbers — dark and light themes switchable on the fly
LSP integration — code completion, go-to-definition, live diagnostics
Language profiles (.prf) — compiler flags, libraries, pkg-config packages per project
Run modes: Auto / Makefile / Custom command per language
MD/TXT Preview panel
File tree — project navigation with Collapse/Expand All
Search in files — full project text search
Two embedded terminals: mshell and bash — always available side by side
Output and Tasks panels — script results and running jobs, alongside mshell and bash terminals
mshell connected via IPC — inline LLM commands, urlcache, eval math, pythonexec, and direct launch of mshell built-in GUI applications (Studio, Notebook, edi) without leaving mide

AI:
Explain / Fix / Generate — inline LLM actions on editor code, result inserted directly
AI Chat — separate chat window: free-form conversation, role loading (.role), prompt templates (.prm), context management
Chat → editor: insert code block, run as script, send to edi as .md workflow — one click
Image generation via prompt, profile-based (size, quality, count)
Image analysis
LLM provider cycling: OllamaN — hundreds of Ollama models (local & cloud), Claude, OpenAI, Gemini

mshell Workflow:
Workflows are plain Markdown (.md) files — human-readable and executable simultaneously, no separate build system
Each fenced code block is a pipeline stage — any of 8 languages compiled and run automatically, data flows via named variables between stages
LLM directives inline — up to 3 models as pipeline stages, async parallel execution with await barriers
Full control flow: WHILE, FOREACH, LOOP, TRY/CATCH, SPLIT/MERGE, IF/routing
AI Chat generates workflows directly
Specialized LLM roles: mshell expert (scripts + workflows), mshell workflow expert (polyglot pipelines), matplotlib expert (data visualization workflows) — loaded on demand from .role files
Generative → Materialized: LLM-driven workflows convert to deterministic, LLM-free production artifacts
Ideal for AI agents — the Intelligent Email Agent for IMAP/POP3 servers is built on mide and mshell Workflow

mshell Ecosystem Integration:
edi — mshell Workflow editor
mshell Studio — Library Manager and Knowledge Base/Validator
mshell Notebook

Git Integration:
Force Sync → GitHub in one click
AI-generated README.md

Tech stack: C · GTK4 · GtkSourceView · VTE · WebKitGTK · mshell IPC
Platform: Linux x86_64 — Ubuntu 22.04 / 24.04 LTS, Debian 12/13
Art2Dec SoftLab

#mshell #mide #IDE #AI #LLM #GTK4 #Linux #Polyglot #mshellWorkflow #Art2DecSoftLab #DevTools #CodeGeneration

MSHELL — a next-generation Unix shell for AI-assisted development and mathematics.

root — Sat, 30 May 2026 22:56:17 +0000

Meet MSHELL — a next-generation Unix shell for AI-assisted development and mathematics by Art2Dec SoftLab. A single C binary that replaces bash as your login shell — adding native LLM integration, polyglot code execution, and a complete Workflow engine without sacrificing standard shell compatibility.

Shell & Scripting:
Bash compatibility — pipes, redirections, standard Unix commands work as-is
Native .ms scripts (#!/bin/mshell) — variables, IF/FOR/WHILE/FUNC, full control flow with AST-based parser
Embedded Lua engine — inline math: arithmetic, trigonometry, statistics, calculus, matrix and vector operations, all via eval
pythonexec — execute Python code inline, one-liners or multi-statement
Native execution of C, C++, Go, Rust, Python, Lua, Bash — compiled and run on demand

AI Integration — Different AI Models Simultaneously:
ollamaN — chat mode: conversational LLM with 200 message pairs persistent context per model
ollamaNexec — exec mode: generate and automatically execute code, bash scripts, Python — no manual steps
-url= / url= option — fetch any web page and pass to LLM in one command
PDF, DOCX, XLSX, DjVu — auto-extracted and included in LLM context when referenced by filename
urlcache — 50-slot web page cache: load, view, refresh, merge slots for multi-source analysis
Pipe mode (-p) — chain LLM output into standard Unix pipes
Image analysis — image=path.png sends image directly to multimodal LLM
Vendors: Ollama (hundreds of local & cloud models), Claude, OpenAI, Gemini

Visualization — Live GUI from CLI:
ollamaNexec commands generate and runs matplotlib, plotly, seaborn and other frameworks code on the fly — chart opens as real GUI window instantly

mshell Workflow:
Plain Markdown (.md) files as executable polyglot pipelines — documentation and code in one file
8 languages per pipeline — bash, python, c, cpp, rust, go, lua, mshell — each block auto compiled and run
Inter-language variable system (Interlang) — named variables flow between any language blocks via MSH_VAR_* files
LLM directives inline () — up to 3 models as pipeline stages, async parallel execution
Full control flow: WHILE, FOREACH, LOOP, TRY/CATCH, SPLIT/MERGE, IF/routing
Generative → Materialized: LLM-driven workflows captured into deterministic, LLM-free production artifacts
Built-in GUI Apps:
edi — visual Workflow and languages editor
mshell Studio — Library Manager (per-language compilation profiles) + Knowledge Base/Validator (error pattern learning, auto-fix)

Platform: Linux x86_64 · ARM64 — Ubuntu 22.04/24.04 LTS, Debian 12/13, Raspberry Pi 4/5
Tech stack: C · embedded Lua · GCC/G++/rustc/go/python3

Art2Dec SoftLab · mshell Ecosystem: [https://www.appservgrid.com/paw92/index.php/2026/02/04/mshell-v-1-4-1-cheatsheet-january-26th-2026/]

#mshell #AI #LLM #Linux #Shell #Polyglot #mshellWorkflow #RaspberryPi #Ollama #Claude #OpenAI #Art2DecSoftLab #DevTools #Mathematics #Automation