Coolify shipped a first-party MCP server in v4.1.0, so I wired Claude Code to the VPS that runs this blog and deployed by asking. It works, which is the worrying part. What the tools actually do (no rollback), the thin confirmation model, and how I scope tokens so an agent can't restart prod off a misread sentence.
The first of my GitHub Trending roundups: the repos actually worth cloning this week — agent scaffolding, from-scratch language runtimes in Rust and C, and two tools that just fix something today — plus how to spot the star-farmed shell hiding in the pile.
OpenCode is my backup for when Claude Code taps out. A field note on provider-agnosticism — and why Copilot going token-based knocked it out of my daily-driver slot.
I pointed Upstash's Context7 MCP at Symfony, Laravel and Drupal contrib churn to see if its injected docs are current, complete, and worth the token budget.
Field notes on connecting a coding agent to a running Drupal install through the AI module's MCP server, and the governance gaps of letting it build content types and run actions.
Field notes on Supacode, a native macOS app that runs many CLI coding agents in parallel, each in its own git worktree. What it does, and the caveats.
An honest daily-driver review of cmux, the open-source native macOS terminal built for running Claude Code and other AI coding agents in parallel.
A local-first, Apache-2.0 app that turns Claude Code, Cursor and 20+ other CLIs into a design tool. What it actually does, and the honest caveats.
Field notes on Keygraph's Shannon: an open-source, white-box AI pentester that reads your source and actually exploits the bugs it finds. What works, what to fear.
A Drupal dev tries AI pull-request review: what CodeRabbit does well, why it gets noisy, the pricing, and whether it belongs on your merge requests.
