added prompt compression agent

agent.py

@@ -6,13 +6,12 @@ import numpy as np
 import google.generativeai as genai
 from dotenv import load_dotenv
 from sentence_transformers import SentenceTransformer
+from transformers import pipeline  # added for summarization
 
 class CodingAgent:
     def __init__(self):
-        
         load_dotenv()
 
-        
         self.api_key = os.getenv("GEMINI_API_KEY")
         if not self.api_key:
             raise ValueError("GEMINI_API_KEY not found in environment or .env file.")
@@ -20,17 +19,17 @@ class CodingAgent:
         genai.configure(api_key=self.api_key)
         self.model = genai.GenerativeModel("gemini-1.5-flash")
 
-        
         self.embedder = SentenceTransformer("all-MiniLM-L6-v2")
         self.index = faiss.IndexFlatL2(384)
         self.docs = []
 
-        
         self.conn = sqlite3.connect("memory.db", check_same_thread=False)
         self.conn.execute(
             """CREATE TABLE IF NOT EXISTS memory (id INTEGER PRIMARY KEY, query TEXT, response TEXT)"""
         )
 
+        self.summarizer = pipeline("summarization", model="facebook/bart-large-cnn")  # added
+
     def embed_chunks(self, texts):
         return self.embedder.encode(texts, convert_to_numpy=True)
 
@@ -67,6 +66,13 @@ class CodingAgent:
         D, I = self.index.search(np.array([query_emb]), top_k)
         return "\n\n".join(self.docs[i] for i in I[0])
 
+    def compress_context(self, context, token_limit=2000):
+        """Summarizes context if it exceeds token limit."""
+        if len(context.split()) < token_limit:
+            return context
+        summary = self.summarizer(context, max_length=200, min_length=50, do_sample=False)[0]['summary_text']
+        return summary
+
     def answer(self, query):
         # Check memory first
         cursor = self.conn.execute(
@@ -77,9 +83,11 @@ class CodingAgent:
             return f"[From memory] {result[0]}"
 
         context = self.retrieve_context(query)
+        compressed_context = self.compress_context(context)
+
         prompt = (
             f"You are a helpful coding assistant.\n\n"
-            f"Context (from uploaded docs):\n{context}\n\n"
+            f"Context (from uploaded docs):\n{compressed_context}\n\n"
             f"User question: {query}\n\n"
             f"Answer with code or explanation where needed."
         )