Changes

Dockerfile

@@ -2,17 +2,16 @@ FROM python:3.12-slim
 
 WORKDIR /app
 
-# install uv (fast dependency manager)
-RUN pip install --no-cache-dir uv
+COPY requirements.txt .
+RUN pip install --no-cache-dir -r requirements.txt
+
 
-# copy everything
 COPY . .
 
-# install project from pyproject.toml
-RUN uv pip install --system -e ".[dev]"
+ENV API_BASE_URL="https://router.huggingface.co/v1"
+ENV MODEL_NAME="Qwen/Qwen2.5-72B-Instruct"
+ENV HF_TOKEN=""
 
-# expose API port (FastAPI default)
 EXPOSE 8000
 
-# run using your pyproject entrypoint
-CMD ["server"]
+CMD ["python", "inference.py"]

requirements.txt

@@ -1,6 +1,6 @@
-fastapi==0.110.0
-uvicorn==0.29.0
-pydantic==2.6.4
+fastapi>=0.110.0
+uvicorn>=0.29.0
 openai>=1.0.0
+python-dotenv>=1.0.0
 openenv-core>=0.2.0
-pyyaml>=6.0
+pydantic>=2.0.0

server/app.py

@@ -5,35 +5,46 @@ Exposes:
   POST /reset
   POST /step
   GET  /state
+  GET  /health
 """
 
 from __future__ import annotations
 from typing import Dict, Any, Optional
 
 import os
+import logging
 import uvicorn
 from fastapi import FastAPI, HTTPException
 from pydantic import BaseModel
 
-# Import your environments
+
+
+# Import environments
 from tasks.task_simple import create_env as create_simple
 from tasks.task_multi_service import create_env as create_multi
 from tasks.task_critical import create_env as create_critical
 
+# Configure logging
+logging.basicConfig(level=logging.INFO, format="%(asctime)s [%(levelname)s] %(message)s")
 
 app = FastAPI(title="DebugOps AI Environment", version="1.0.0")
 
 # Global env instance
 _env = None
 
+
 class ResetRequest(BaseModel):
+    """Request schema for resetting the environment."""
     task: str = "simple"   # simple | multi_service | critical
 
 
 class StepRequest(BaseModel):
+    """Request schema for stepping through the environment."""
     action: str
 
+
 def create_env(task: str):
+    """Factory method to create environment based on task type."""
     if task == "simple":
         return create_simple()
     elif task == "multi_service":
@@ -45,13 +56,16 @@ def create_env(task: str):
 
 
 def get_env():
+    """Retrieve the current environment instance."""
     global _env
     if _env is None:
         raise HTTPException(status_code=400, detail="Call /reset first")
     return _env
 
+
 @app.get("/")
 def root():
+    """Root endpoint providing metadata and available endpoints."""
     return {
         "name": "DebugOps AI Environment",
         "description": "Production debugging RL environment (OpenEnv compatible)",
@@ -61,14 +75,17 @@ def root():
 
 @app.get("/health")
 def health():
+    """Health check endpoint."""
     return {"status": "ok"}
 
 
 @app.post("/reset")
 def reset(request: Optional[ResetRequest] = None):
+    """Reset the environment with a given task."""
     global _env
     try:
         task = request.task if request else "simple"
+        logging.info(f"Resetting environment with task: {task}")
 
         _env = create_env(task)
         obs = _env.reset()
@@ -79,15 +96,18 @@ def reset(request: Optional[ResetRequest] = None):
         }
 
     except Exception as e:
+        logging.error(f"Error during reset: {e}")
         raise HTTPException(status_code=500, detail=str(e))
 
 
 @app.post("/step")
 def step(request: StepRequest):
+    """Perform a step in the environment with the given action."""
     env = get_env()
 
     try:
         obs, reward, done, info = env.step(request.action)
+        logging.info(f"Step taken: {request.action}, reward={reward}, done={done}")
 
         return {
             "observation": obs,
@@ -97,19 +117,28 @@ def step(request: StepRequest):
         }
 
     except Exception as e:
+        logging.error(f"Error during step: {e}")
         raise HTTPException(status_code=400, detail=str(e))
 
 
 @app.get("/state")
 def state():
+    """Retrieve the current environment state."""
     env = get_env()
     try:
         return env.state()
     except Exception as e:
+        logging.error(f"Error retrieving state: {e}")
         raise HTTPException(status_code=400, detail=str(e))
 
+
+
+
+
 def main():
+    """Entry point for running the FastAPI server."""
     port = int(os.getenv("PORT", 7860))
+    logging.info(f"Starting DebugOps server on port {port}")
     uvicorn.run(app, host="0.0.0.0", port=port)
 
 

ui.py

@@ -1,100 +0,0 @@
-import os
-
-import gradio as gr
-import requests
-
-BASE_URL = os.getenv("SPACE_URL", "http://localhost:8000")
-
-current_logs = []
-current_actions = []
-
-
-def reset_env(task):
-    try:
-        res = requests.post(f"{BASE_URL}/reset", json={"task": task})
-        res.raise_for_status()
-
-        data = res.json()
-        obs = data["observation"]
-
-        return (
-            format_logs(obs.get("logs", [])),
-            "Environment reset",
-            "Running...",
-        )
-
-    except Exception as e:
-        return ("Error occurred", str(e), "Failed", None)
-
-
-def step_env(action):
-    try:
-        res = requests.post(f"{BASE_URL}/step", json={"action": action})
-        res.raise_for_status()
-
-        data = res.json()
-
-        obs = data["observation"]
-        reward = data["reward"]
-        done = data["done"]
-
-        return (
-            format_logs(obs.get("logs", [])),
-            f"{action} → {reward}",
-            "Done" if done else "Running",
-        )
-
-    except Exception as e:
-        return ("Error occurred", str(e), "Failed", None)
-
-
-def format_logs(logs):
-    return "\n".join([f"• {log}" for log in logs])
-
-
-def build_ui():
-    with gr.Blocks() as demo:
-        gr.Markdown("#DebugOps AI Environment")
-        gr.Markdown("Interactive incident response simulator")
-
-        task_dropdown = gr.Dropdown(
-            ["simple", "multi_service", "critical"],
-            label="Select Task",
-            value="simple"
-        )
-
-        reset_btn = gr.Button("Reset Environment")
-
-        logs_box = gr.Textbox(label="Logs", lines=10)
-        actions_box = gr.Textbox(label="Actions Taken", lines=10)
-        status_box = gr.Textbox(label="Status")
-
-        with gr.Row():
-            gr.Button("Restart API").click(
-                lambda: step_env("restart_api"),
-                outputs=[logs_box, actions_box, status_box]
-            )
-            gr.Button("Restart DB").click(
-                lambda: step_env("restart_db"),
-                outputs=[logs_box, actions_box, status_box]
-            )
-            gr.Button("Restart Cache").click(
-                lambda: step_env("restart_cache"),
-                outputs=[logs_box, actions_box, status_box]
-            )
-            gr.Button("Scale Up").click(
-                lambda: step_env("scale_up"),
-                outputs=[logs_box, actions_box, status_box]
-            )
-            gr.Button("No-op").click(
-                lambda: step_env("noop"),
-                outputs=[logs_box, actions_box, status_box]
-            )
-
-        reset_btn.click(
-            reset_env,
-            inputs=task_dropdown,
-            outputs=[logs_box, actions_box, status_box]
-        )
-
-    return demo