point refs to final public homes + drop placeholder

Dockerfile

@@ -21,7 +21,7 @@ RUN if [ -x /app/llama-server ]; then ln -sf /app/llama-server /usr/local/bin/ll
     else ln -sf "$(find / -name llama-server -type f 2>/dev/null | head -1)" /usr/local/bin/llama-server; fi \
     && llama-server --version 2>&1 | head -1
 
-# The GGUF is pulled at RUNTIME by app.py from the PRIVATE model repo using the HF_TOKEN Space secret
+# The GGUF is pulled at RUNTIME by app.py from the (now public) model repo - HF_TOKEN optional
 # (MODEL_REPO/MODEL_FILE below) - not baked into the image, so the private model never lives in the layers.
 
 # App code + bundled backend / schema. The tokenizer is NOT bundled - app.py pulls it at runtime from the
@@ -37,10 +37,10 @@ ENV CODEAGENT_PROJ=/app \
     CODEAGENT_THREADS=2 \
     CODEAGENT_MAX_ITERS=20 \
     CODEAGENT_BASH_TIMEOUT=150 \
-    MODEL_REPO=WeReCooking2/MiniCPM5-1B-Agent-GGUF \
+    MODEL_REPO=Luminia/MiniCPM5-1B-Agent-GGUF \
     MODEL_FILE=MiniCPM5-1B-Agent-v4-Q8_0.gguf \
     HF_HOME=/app/.hf
-# HF_TOKEN injected at runtime as a Space SECRET (to read the private model repo).
+# HF_TOKEN optional now the model repo is public (kept for back-compat; popped after download).
 
 # Override the llama.cpp image's entrypoint - we launch the Python app, which spawns llama-server itself.
 ENTRYPOINT []

README.md

@@ -27,7 +27,6 @@ Q8_0 GGUF on llama.cpp, no GPU. Give it a task; it reasons in `<think>`, then us
 the chat. Multi-turn: files and history persist across messages. It is also exposed as an **MCP tool**
 (`run_coding_task` at `/gradio_api/mcp/`).
 
-▶️ **Demo video:** `<PASTE_VIDEO_LINK>`  ·  📣 **Social post:** `<PASTE_DISCORD_POST_LINK>`
 
 ## What it is
 
@@ -58,7 +57,9 @@ shows it working end-to-end, so it can be judged even if a live run is slow).
 
 ## Model, dataset & full reproduction
 
-→ **[WeReCooking2/MiniCPM5-1B-Agent-GGUF](https://huggingface.co/WeReCooking2/MiniCPM5-1B-Agent-GGUF)** (model card =
+→ **[Luminia/MiniCPM5-1B-Agent-GGUF](https://huggingface.co/Luminia/MiniCPM5-1B-Agent-GGUF)** (model card =
 the full data mix, SFT/DPO recipe, eval, and exact reproduce commands; v4 dataset bundled under `dataset/`).
 
-*Built for the Build Small Hackathon · OpenBMB category.*
+💻 **Code on GitHub:** [Katehuuh/MiniCPM5-1B-Agent](https://github.com/Katehuuh/MiniCPM5-1B-Agent) (the Space + the full training pipeline; code reviewed with OpenAI Codex).
+
+*Built for the Build Small Hackathon · OpenBMB + OpenAI Codex tracks.*

app.py

@@ -976,7 +976,7 @@ WELCOME = (
     "run it in a sandbox, read the output, debug, and show the result inline.\n\n"
     "📊 charts & images  ·  🖱️ little web apps  ·  🧮 compute & web look-ups"
 )
-_MODEL_REPO = os.environ.get("MODEL_REPO", "WeReCooking2/MiniCPM5-1B-Agent-GGUF")
+_MODEL_REPO = os.environ.get("MODEL_REPO", "Luminia/MiniCPM5-1B-Agent-GGUF")
 FOOTER_HTML = (
     '<div id="appfooter">'
     '<span class="ft-mark">&#9613;</span>'