@runcore-sh/runcore 0.5.7 → 0.5.9

package/README.md

@@ -24,8 +24,8 @@ If you've ever watched an agent burn tokens repeating the same mistake three tim
 
 ```bash
 # 1. Clone
-git clone https://github.com/XDM-ZSBW/core.git
-cd core
+git clone https://github.com/XDM-ZSBW/runcore.git
+cd runcore
 
 # 2. Install
 npm install
@@ -345,6 +345,7 @@ The key innovation is not autonomy — many systems are autonomous. It's **adapt
 - **[The File System Is the New Database: How I Built a Personal OS for AI Agents](https://x.com/koylanai/status/2025286163641118915)** — Muratcan Koylan. Personal Brain OS, progressive disclosure, format-function mapping, skill system.
 - [Agent Skills for Context Engineering](https://github.com/muratcankoylan/Agent-Skills-for-Context-Engineering) — Skill format and patterns.
 - [Agent Architecture: From Prompt to Context](https://www.agent32.org/agent-architecture-overview-from-prompt-to-context/) — Context engineering as the core discipline.
+- **[Context Engineering: From Vibe Coder to Software Engineer](https://augmentedadvisors.com/book/)** — Craig Trulove. Practical guide to production AI systems: system prompts, RAG, multi-agent architectures, memory systems, testing, and deployment.
 - [Beyond the Prompt (COALA)](https://medium.com/google-cloud/beyond-the-prompt-why-your-next-ai-agent-needs-a-brain-and-how-coala-research-paper-provides-an-ba187a906ea0) — Modular memory, internal/external actions.
 
 ---

package/dictionary.json

@@ -1,6 +1,6 @@
 {
-  "version": "0.5.7",
-  "publishedAt": "2026-03-11T23:13:12.177Z",
+  "version": "0.5.9",
+  "publishedAt": "2026-03-12T00:06:35.330Z",
   "specs": [],
   "glossary": {
     "brain": "The local file-based data store for an instance. Always local. Never hosted.",

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@runcore-sh/runcore",
-  "version": "0.5.7",
+  "version": "0.5.9",
   "description": "Local-first AI agent runtime. One command, your machine, your data.",
   "type": "module",
   "main": "dist/index.js",
@@ -11,6 +11,7 @@
   "files": [
     "dist",
     "public",
+    "!public/avatar",
     "skills",
     "dictionary.json",
     "brain-template",

package/public/avatar/README.md

@@ -1,43 +0,0 @@
-# Core Avatar
-
-Place a GLB avatar model here as `dash.glb` to enable the 3D TalkingHead avatar.
-
-## How it works
-
-Core uses [TalkingHead.js](https://github.com/met4citizen/TalkingHead) to render a 3D avatar in the chat UI. When Piper TTS speaks, the audio drives lip-sync animations on the model.
-
-If no `dash.glb` file is found, Core falls back to a CSS-animated placeholder.
-
-## Getting a compatible model
-
-The GLB model needs:
-- **Mixamo-compatible skeleton rig** (standard humanoid bone names)
-- **52 ARKit blend shapes** (eyeBlinkLeft, jawOpen, mouthSmile, etc.)
-- **15 Oculus viseme blend shapes** (viseme_sil, viseme_PP, viseme_FF, etc.)
-
-### Option 1: Avaturn (recommended, free)
-
-1. Go to [avaturn.me](https://avaturn.me/)
-2. Create an avatar (photo upload or customize)
-3. Export as GLB with ARKit blend shapes enabled
-4. Save as `public/avatar/dash.glb`
-
-### Option 2: MPFB (Blender, free)
-
-1. Install [MPFB](https://static.makehumancommunity.org/mpfb.html) Blender extension
-2. Create a character, export as GLB
-3. Use the TalkingHead [build-visemes-from-arkit.py](https://github.com/met4citizen/TalkingHead/blob/main/blender/build-visemes-from-arkit.py) Blender script to add Oculus visemes from ARKit shapes
-4. Save as `public/avatar/dash.glb`
-
-### Option 3: Any GLB with the right blend shapes
-
-Any GLB/GLTF model with the required blend shapes will work. The model is loaded via Three.js at `/public/avatar/dash.glb`.
-
-## Customization
-
-The avatar initialization is in the `<script type="module">` block at the bottom of `public/index.html`. You can adjust:
-
-- `cameraView`: `"full"`, `"mid"`, `"upper"`, `"head"`
-- `avatarMood`: `"neutral"`, `"happy"`, `"angry"`, `"sad"`, `"love"`, etc.
-- `body`: `"M"` or `"F"` (affects idle animation style)
-- Lighting: ambient/direct color and intensity