ANTIGRAVITY V9.2.0 ADDS LOCAL HF EVALS, HF JOBS VISION TRAINING, TRANSFORMERS.JS, AND BETTER JQ/TMUX WORKFLOWS
Antigravity-awesome-skills v9.2.0 ships a big Hugging Face-focused update plus serious shell workflow upgrades. The [release notes](https://github.com/sickn33/...
Antigravity-awesome-skills v9.2.0 ships a big Hugging Face-focused update plus serious shell workflow upgrades.
The release notes add skills for local model evals with inspect-ai and lighteval, Gradio app editing, Trackio experiment tracking, and vision training on Hugging Face Jobs. It also introduces Transformers.js support for JS/TS inference.
For day-to-day ops, it folds in deep jq and tmux guidance, and tighter Odoo guardrails for credentials and EDI reliability. Everything targets Claude Code, Cursor, Codex CLI, Gemini CLI, and similar assistant workflows.
Gives teams a ready-made path to standardize local model evals and Hugging Face training flows.
Improves reliability of terminal-first pipelines and Odoo-backed integrations without writing custom playbooks.
-
terminal
Run a small eval suite locally with inspect-ai/lighteval and compare results against your current benchmarking workflow.
-
terminal
Train a toy vision model on Hugging Face Jobs and validate logging/alerts via Trackio in a CI-triggered run.
Legacy codebase integration strategies...
- 01.
Swap ad-hoc shell snippets with the new jq/tmux workflows and measure reduced pager fatigue and session churn.
- 02.
Pilot the Odoo guidance in a non-prod EDI flow to verify credential scoping and error recovery.
Fresh architecture paradigms...
- 01.
Base a lightweight ML stack on HF Jobs + Trackio, with Gradio for demos and Transformers.js for browser inference.
- 02.
Bake local evals into your scaffolding so every new model change ships with reproducible metrics.