feat: refresh gemini-cli sample with gemini-3.1-pro / 3-flash / 2.5-pro / 2.5-flash

Replaces the deprecated gemini-2.5-flash-lite. Tier mapping:
- defaultModel / lightModel: gemini-2.5-flash
- mediumModel: gemini-3-flash
- heavyModel: gemini-3.1-pro

Author: atomantic

CHANGELOG.md

@@ -1,5 +1,11 @@
 # Changelog
 
+## [0.8.4] - 2026-05-03
+
+### Improvements
+
+- Refresh `gemini-cli` sample provider with the current Google model lineup: `gemini-3.1-pro`, `gemini-3-flash`, `gemini-2.5-pro`, `gemini-2.5-flash` (replaces the now-deprecated `gemini-2.5-flash-lite`). Tier mapping: `defaultModel`/`lightModel` → `gemini-2.5-flash`, `mediumModel` → `gemini-3-flash`, `heavyModel` → `gemini-3.1-pro`.
+
 ## [0.8.3] - 2026-04-25
 
 ### Improvements

package-lock.json

@@ -1,6 +1,6 @@
 {
   "name": "portos-ai-toolkit",
-  "version": "0.8.3",
+  "version": "0.8.4",
   "description": "Shared AI provider, model, and prompt template patterns for PortOS-style applications",
   "author": "Adam Eivy <adam@eivy.com> (https://atomantic.com)",
   "license": "MIT",

package.json

@@ -60,8 +60,11 @@
       "type": "cli",
       "command": "gemini",
       "args": [],
-      "models": [],
-      "defaultModel": null,
+      "models": ["gemini-3.1-pro", "gemini-3-flash", "gemini-2.5-pro", "gemini-2.5-flash"],
+      "defaultModel": "gemini-2.5-flash",
+      "lightModel": "gemini-2.5-flash",
+      "mediumModel": "gemini-3-flash",
+      "heavyModel": "gemini-3.1-pro",
       "timeout": 300000,
       "enabled": true,
       "envVars": {},