We removed an unauthenticated debug route, fixed a license-check bypass, floored 12 pip CVEs and made CI build the app before merge. Plus layout, chat and reviewer fixes.
Mastra 1.79 went to production and broke Maestro's working memory with a TLS error. We fixed it forward the same day, then verified RDS TLS on every Postgres client in the repo.
Three model-registry PRs so new model families and their prices land automatically, the PR reviewer re-reviews pushed fixes, and admin flags failed signups and redacts tokens in bounces.
This week we applied three prompt audits across the ChameleonLabs fleet, retired an agent nobody used anymore, and worked through 98 Snyk advisories in ImageAI.
Seven gallery PRs on LelandGreenProductions (art gallery, owner Studio, Le Bot, albums, a mobile fix, AI cost audit) plus an automated docs refresh on the ChameleonLabs site.
Four PRs this week: a full Sprite CLI for ImageAI (v0.49.0), a 23-finding auth and chat cleanup in the ChameleonLabs app, a tilde bug in our PR reviewer and an automated docs refresh.
Four PRs this week: the ChameleonLabs site now pulls prices from Stripe instead of hand-typed copy, the September prompt audit landed in Maestro, and ImageAI got a Sprite tab.
We merged ImageAI #42: a DataPaths resolver, four relocatable storage groups (Images, Video, Models, Logs), a Storage Locations settings panel and a safe migration path.
We restyled about 50 of Maestro's chat components to the Claude Desktop structural recipe: semantic tokens, border elevation and roomier typography. The cyan and navy didn't change.
This week: server-side Turnstile and rate limits on auth, a Discord /feedback pipeline into a new suggestions table, a quieter ChatMaster, five new plugins and an ImageAI parameter fix.
Humanizer went live for ChameleonLabs members this week. ImageAI picked up Custom Styles and curved overlay text, and ChatMaster's reports were improved.
Nine PRs merged: ImageAI gained comic-style page layouts, a layout CLI and Gemini Omni video with conversational editing, and we hardened the agent system behind ChameleonLabs.
QuickStock, Tincture, Maestro, and our agent fleet now report LLM token cost into one per-app rollup. Plus a Discord approval gate for our agents and a self-updating model registry.