Andrea Grandi

@andreagrandi.it

Software Developer at SuperTab - Python, Go mentor at Exercism - 🏳️‍🌈 🇮🇹 🇪🇺 🤝 🇺🇦 🇵🇸

GPT Sol 6 and Luna 6 released and they cost 50% less than their 5.6 equivalents! Yay!? No. I'm not excited. Give them a few days and they will silently and shadly decrease the available usage and people will start complaining the week only last 3 days. See you in 1-2 weeks.

If you are hammering your Codex subscription with GPT-5.6-sol (and complaining about usage) only, you are probably doing it wrong. Give sol high for planning + luna xhigh for implementation + sol xhigh for review a try. You can easily split these tasks if you use OhMyPi. You are welcome!

I'm seriously considering cancelling my ChatGPT subscription after this shit show community.openai.com/t/codex-limi... TL/DR: usage is draining faster than ever. People are exhausting 20x subscriptions in a couple of days. OpenAI is ignoring the complains, censoring posts and not ack the issue.

Codex limits spark frustration among subscribers

I’m facing the exact same issue, and this is honestly extremely frustrating. I purchased the Pro 20× plan (~₹16,000) specifically because I do very heavy development work and needed the higher Codex ...

community.openai.com

Working on open source apps has become a quite expensive hobby (assuming you write software like it's 2026 not like it's < 2024). While I can justify a ~$20 AI subscription for an hobby, I can hardly justify a ~$100 one (unless you make money from it). Enjoy what I can provide with my $20 🤷🏻‍♂️

Released mcp-wire 0.3.0 🎉 A Go CLI that configures MCP servers across Claude Code, Codex, and OpenCode from one interface. New: doctor command, richer TUI metadata, machine-readable metadata, GitHub/Notion/Linear services, safer credential storage. www.andreagrandi.it/posts/announ...

Announcing mcp-wire 0.3.0

mcp-wire 0.3.0 is out with a new doctor command, richer TUI metadata, machine-readable metadata for agents, and more curated MCP services including GitHub, Notion, and Linear.

andreagrandi.it

logbasset already documents itself for AI agents via a `context` command. So why add an agent skill too? Because self-documentation ≠ discovery: one tells the agent how to use the tool, the other gets it to reach for the tool at all. www.andreagrandi.it/posts/why-ad...

Why add an agent skill to a CLI that already has a context command?

logbasset already self-documents for agents through a context command, so why add an installable agent skill on top? Because self-documentation and discovery are two different problems.

andreagrandi.it

💛 PyCon Italia 2026 has come to an end. Thank you to everyone who made these days unforgettable. Speakers, volunteers, sponsors, organizers, partners, and attendees. You are the reason this community feels like home. 🐍 We'll be back in Bologna on June 2–5, 2027. 👋 See you next year! #PyConIT

A large group of PyCon Italia attendees, volunteers, and organizers gathered outdoors near a pool, waving at the camera during the closing group photo of the conference.

🚀 Sentire v0.4.0 — a CLI for the Sentry API, built for fast incident triage in your terminal. New: • 🏠 Self-hosted Sentry (SENTRY_API_BASE_URL) • ⚡ Priority + relative times in issue lists • 🔒 API token redaction in errors github.com/andreagrandi... #golang #sentry #cli

GitHub - andreagrandi/sentire: Command Line Tool for Sentry API

Command Line Tool for Sentry API. Contribute to andreagrandi/sentire development by creating an account on GitHub.

github.com