weiland

Software architect and metal fabricator. Ex Zamboni driver.

[weiland @weiland]

@AetherRev @mrfokki I asked Claude and it didn't know and said I should do it right then and there just to be safe. I showed it my normal expire time of 11pm and it said I should be safe till 7pm so I went with that as I was only 12% in. NOPE.

[weiland @weiland]

@antigravity Nevermind, I inadvertently set it to Claude Opus NOT Flash 3.8 which I thought so I burned through Claude 5 hour limit in Antigravity so quickly, which is on par for the course...

[weiland @weiland]

There's a 1st, I hit the 5 hour @antigravity limit in under 1 hour. I was at ~72% Argon usage this morning, I had it fix it's PR it went down to 57%. Started a new session with 3.8 Flash (Medium) and bam I'm out of usage for the next 4 hours.

[weiland @weiland]

@oldstackjournal I use Antigravity to orchestra long tasks that will affect multiple repos (API & UI) and Herdr. In herdr each repo has panes for Opus, Sonnet, AGY. In Antigravity I tell it what pane to use for each subagent to use (dev Opus, test AGY, Reviewer Sonnet).

[weiland @weiland]

@oldstackjournal Weekends and evenings I'm slowly updating and adding new features to https://t.co/GjVZNoCGfU and then design and fabricate metal trivets https://t.co/YEWZxckhrY Solution Architect by day.

[weiland @weiland]

@ProductHunt Version 5.1 of https://t.co/GjVZNoCGfU, which was 25 years in the making. It was finally time to update how certificate assets were added to certificates.

[weiland @weiland]

What’s next: I'll be expanding @mattpocockuk’s /implement with pre/post-hook lifecycle events (like our multi-agent /goal loops) so the agent becomes self-learning across prior retros.

CC 5.1 is officially live! Walkthrough & blog post: https://t.co/UxkBLmth30 (6/6)

[weiland @weiland]

Last weekend I used @antigravity IDE and @mattpocockuk /grill-me-with-docs skill to flesh out a major Component Gallery overhaul for https://t.co/6uMk3hUw7M

Before writing a single line of code, the Socratic grilling forced me to resolve every UX edge case and commit ADRs. (1/6)

[weiland @weiland]

During pre-merge review with my /ruthless-code-review skill, the agent caught a critical landmine: our new GA4 telemetry was transmitting saved certificate titles (which contain student/employee names).

Caught and sanitized before a single byte of PII hit production. (5/6)

[weiland @weiland]

@jamonholmgren I do similar for my /goal runs, I have a file with my best practices that gets loaded on a prehook. A posthook does a goal retro and writes out the postmortem for the goal run and if there are any lessons that need to be used go into the prehook file.

[weiland @weiland]

@rodydavis When using /goal in agy 2.0/IDE I want to specify what model the subagents need to use. Right now all subagents inherit whatever model was set when I started the orchestration.

[weiland @weiland]

Been focusing on https://t.co/v0GtMMADxS the last few months. Had some https://t.co/EDhvrwgdNz orders come in this week so will switch gears and break out the welder. New design dropping too!

[Certificate Creator @CertCreator]

Want to kick the tires before you commit any money, https://t.co/e53pdwVzpg now has a free account. Prints are watermarked or start with the $2.99/mo plan no watermarks, unlimited prints and upload your logo or images to be used.

[weiland @weiland]

@DannPetty How about a way to see API creations in the Quiver UI? I can see the calls but seeing the end SVG would be helpful.

[weiland @weiland]

@GoogleCloudTech In @antigravity IDE I wanted the orchestrator to use 4 different models for each agent. The first run it ran, but when I attempted to use it again it came clean and said it could only use the 1 model for each player in the orchestration, multi models not supported

[weiland @weiland]

@DannPetty Here's the Rampant Lion @QuiverAI created for me.

[weiland @weiland]

@mattpocockuk My vault is a separate repo. Skills are symlinked to each of my project repos. Memory.md is written back into the vault under the project. When I run a goal or workflow it grabs the lessons learned & applies that intelligence. Then writes back lessons learned for next time.