The Elyra blog

The Release With No New Features (And Why That's a Good Thing) - Elyra v0.9.11

Elyra v0.9.11 ships zero new features and deletes 143 any types from @elyracode/web-ui. Here's why a types-only release is some of the most valuable work we do — with real before-and-after from the diff.

SQLite Grew Up: Meet Elyra SQL Anywhere

Every developer has a soft spot for SQLite. It's the database that just works — one file, no server, no config, no 3 a.m. pages about a connection pool. It's in your phone, your browser, your tests, probably your toaster. But there's always been a moment, usually around the time your app finds its second user, when you sigh and say: "Well, time to move to Postgres." What if you didn't have to?

When to Use Elyra: The Agent Harness, Explained

You can talk to a model directly — the same way you can drive a car by turning the alternator with your hand. The model is the engine; the agent harness is everything that turns it into something you'd actually want to drive. Here's what a harness is, why it matters more than which model you picked this week, and when reaching for Elyra is the right call.

Is GLM 5.2 Real, or Just This Month's Hype? (And How to Try It in Elyra in 60 Seconds)

GLM 5.2 from Z.ai is all over your timeline. Here's the un-fun version: what's measurable (a 1M context window, the price), what's vibes (leaderboards), and how to run it in Elyra across five providers in about a minute — plus the routing trick that actually beats the hype.

PHP Gets a Real Brain, and Svelte Joins the Laravel Party - Elyra v0.9.10

Elyra and Git Worktrees Explained: The Quiet Superpower for Running Agents in Parallel

Running more than one AI agent at once means they'd collide — one working directory, one branch, instant chaos. Git worktrees give each agent its own files and branch over a shared history, and Elyra is built to understand the branch, session, and memory of each one independently.

Your Terminal Has Amnesia. We Gave It a Memory

Terminals forget everything the moment you close them. Elyra Conductor v0.8.1 persists every command — the real command line, exit code, duration, and a tail of output — to a tiny local SQLite database, then makes it searchable across every session. Find that fix from three weeks ago by searching the symptom you remember, not the command you forgot.

Stop Babysitting One Agent. Conduct a Whole Orchestra

Elyra Conductor v0.8.0 turns the single-agent workflow into something you can conduct — isolated worktrees, at-a-glance agent presence, and inline PR status, so one person can run several agents instead of babysitting one.

/tree: The Command That Makes Your Agent a Time Machine, Not a Typewriter

Most AI coding agents work like a typewriter — one line of text marching forward, no way back. Elyra stores your session as a tree, and /tree is the map: every rewind, fork, and replay branches off history instead of overwriting it. Exploration becomes free.

The audit was wrong (and that's the point)

Pointing a strong model at your own instruction files to find the rot is a genuinely good idea. We ran it on Elyra's own rulebook — and learned that a confident finding and a correct finding look identical until you check. Here's exactly what happened, false starts included.