[Chandigarh] Aashish — RAG-to-MCP Submission

uc-0a/agents.md

@@ -1,27 +1,15 @@
-# agents.md — UC-0A Complaint Classifier
-# INSTRUCTIONS:
-# 1. Open your AI tool
-# 2. Paste the full contents of uc-0a/README.md
-# 3. Use this prompt:
-#    "Read this UC README. Using the R.I.C.E framework, generate an
-#     agents.md YAML with four fields: role, intent, context, enforcement.
-#     Enforcement must include every rule listed under
-#     'Enforcement Rules Your agents.md Must Include'.
-#     Output only valid YAML."
-# 4. Paste the output below
-
 role: >
-  [FILL IN]
+  An AI classifier reading each complaint for the City Operations team to output a category, priority, reason, and flag.
 
 intent: >
-  [FILL IN]
+  To accurately classify complaints, avoiding false confidence on ambiguity, missing justifications, or hallucinated categories, so the output can feed the Director's dashboard every Monday.
 
 context: >
-  [FILL IN]
+  The City Operations team receives hundreds of complaints per week. A naive prompt produces a classifier that invents category names, misses urgent complaints involving children and injuries, and gives confident answers on genuinely ambiguous inputs.
 
 enforcement:
-  - "[FILL IN: category enum rule]"
-  - "[FILL IN: severity keyword rule — list the keywords]"
-  - "[FILL IN: reason field rule]"
-  - "[FILL IN: ambiguity refusal rule]"
-  - "[FILL IN: no invented categories rule]"
+  - "Category must be exactly one value from the allowed list: Pothole, Flooding, Streetlight, Waste, Noise, Road Damage, Heritage Damage, Heat Hazard, Drain Blockage, Other. No variations."
+  - "Priority must be Urgent if description contains any severity keyword: injury, child, school, hospital, ambulance, fire, hazard, fell, collapse."
+  - "Every output row must include a reason field citing specific words from the description."
+  - "If category cannot be determined confidently — output `category: Other` and `flag: NEEDS_REVIEW`."
+  - "Never invent category names outside the allowed list."

uc-0a/classifier.py

@@ -10,21 +10,128 @@
 """
 import argparse
 import csv
+import sys
+import re
+
+ALLOWED_CATEGORIES = {
+    "Pothole", "Flooding", "Streetlight", "Waste", "Noise", 
+    "Road Damage", "Heritage Damage", "Heat Hazard", "Drain Blockage", "Other"
+}
+
+SEVERITY_KEYWORDS = ["injury", "child", "school", "hospital", "ambulance", "fire", "hazard", "fell", "collapse"]
+
+CATEGORY_KEYWORDS = {
+    "Pothole": ["pothole", "crater"],
+    "Flooding": ["flood", "waterlog", "water logged"],
+    "Streetlight": ["street light", "streetlight", "dark", "no light"],
+    "Waste": ["garbage", "waste", "trash", "rubbish"],
+    "Noise": ["noise", "loud", "music", "party", "sound"],
+    "Road Damage": ["crack", "road damage", "broken road", "sinkhole"],
+    "Heritage Damage": ["heritage", "monument", "statue", "ruins"],
+    "Heat Hazard": ["heat", "sun", "blazing", "heatwave"],
+    "Drain Blockage": ["drain", "clog", "choke", "blockage"],
+}
 
 def classify_complaint(row: dict) -> dict:
     """
-    Classify a single complaint row.
+    Classify a single complaint row based on enforcement rules.
     Returns dict with: complaint_id, category, priority, reason, flag
     """
-    raise NotImplementedError("Build this using your AI tool + agents.md")
+    description = row.get("description", "").lower()
+
+    # Error handling: vague/short descriptions
+    if len(description.split()) < 3:
+        return {
+            "complaint_id": row.get("complaint_id", row.get("id", "")),
+            "category": "Other",
+            "priority": "Low",
+            "reason": "Vague or very short description.",
+            "flag": "NEEDS_REVIEW"
+        }
+
+    priority = "Standard"
+    reason_words = []
+
+    # Priority enforcement rule
+    for kw in SEVERITY_KEYWORDS:
+        if re.search(r'\b' + kw + r'\b', description):
+            priority = "Urgent"
+            reason_words.append(kw)
+
+    # Category enforcement rule
+    category = "Other"
+    found_categories = set()
+    for cat, kws in CATEGORY_KEYWORDS.items():
+        for kw in kws:
+            if re.search(r'\b' + kw + r'\b', description):
+                found_categories.add(cat)
+                reason_words.append(kw)
+
+    if len(found_categories) == 1:
+        category = list(found_categories)[0]
+
+    # Ambiguity refusal rule
+    flag = ""
+    if category == "Other" or len(found_categories) > 1:
+        category = "Other"
+        flag = "NEEDS_REVIEW"
+
+    reason_text = "Found relevant keywords: " + ", ".join(set(reason_words)) if reason_words else "No specific keywords mapped."
+
+    return {
+        "complaint_id": row.get("complaint_id", row.get("id", "")),
+        "category": category,
+        "priority": priority,
+        "reason": reason_text,
+        "flag": flag
+    }
 
 def batch_classify(input_path: str, output_path: str):
     """Read input CSV, classify each row, write results CSV."""
-    raise NotImplementedError("Build this using your AI tool + agents.md")
+    results = []
+    try:
+        with open(input_path, 'r', encoding='utf-8') as f:
+            reader = csv.DictReader(f)
+            # Find ID column
+            id_col = None
+            if reader.fieldnames:
+                for col in reader.fieldnames:
+                    if "id" in col.lower():
+                        id_col = col
+                        break
+
+            for row in reader:
+                try:
+                    # Malformed rows logging and skipping
+                    if not row.get('description'):
+                        print(f"Skipping malformed row: {row}")
+                        continue
+
+                    res = classify_complaint(row)
+                    # Use the correct original ID column if available
+                    if id_col:
+                        res["complaint_id"] = row.get(id_col, "")
+                    results.append(res)
+                except Exception as e:
+                    print(f"Error processing row {row}: {e}")
+                    continue
+    except FileNotFoundError:
+        print(f"Input file not found: {input_path}")
+        sys.exit(1)
+
+    if not results:
+        print("No valid rows processed.")
+        return
+
+    fieldnames = ["complaint_id", "category", "priority", "reason", "flag"]
+    with open(output_path, 'w', newline='', encoding='utf-8') as f:
+        writer = csv.DictWriter(f, fieldnames=fieldnames)
+        writer.writeheader()
+        writer.writerows(results)
 
 if __name__ == "__main__":
     parser = argparse.ArgumentParser(description="UC-0A Complaint Classifier")
-    parser.add_argument("--input",  required=True)
+    parser.add_argument("--input", required=True)
     parser.add_argument("--output", required=True)
     args = parser.parse_args()
     batch_classify(args.input, args.output)

uc-0a/results_ahmedabad.csv

@@ -0,0 +1,16 @@
+complaint_id,category,priority,reason,flag
+AM-202401,Other,Standard,No specific keywords mapped.,NEEDS_REVIEW
+AM-202402,Other,Standard,No specific keywords mapped.,NEEDS_REVIEW
+AM-202405,Other,Standard,No specific keywords mapped.,NEEDS_REVIEW
+AM-202406,Heat Hazard,Standard,Found relevant keywords: heatwave,
+AM-202407,Other,Urgent,Found relevant keywords: child,NEEDS_REVIEW
+AM-202410,Pothole,Standard,Found relevant keywords: pothole,
+AM-202414,Other,Standard,No specific keywords mapped.,NEEDS_REVIEW
+AM-202417,Other,Standard,"Found relevant keywords: heritage, waste",NEEDS_REVIEW
+AM-202421,Noise,Standard,Found relevant keywords: music,
+AM-202424,Other,Standard,No specific keywords mapped.,NEEDS_REVIEW
+AM-202429,Other,Standard,No specific keywords mapped.,NEEDS_REVIEW
+AM-202431,Heritage Damage,Standard,Found relevant keywords: heritage,
+AM-202435,Heat Hazard,Standard,Found relevant keywords: heat,
+AM-202444,Waste,Standard,Found relevant keywords: waste,
+AM-202445,Heat Hazard,Standard,Found relevant keywords: sun,

uc-0a/results_hyderabad.csv

@@ -0,0 +1,16 @@
+complaint_id,category,priority,reason,flag
+GH-202401,Other,Urgent,Found relevant keywords: ambulance,NEEDS_REVIEW
+GH-202402,Drain Blockage,Standard,Found relevant keywords: drain,
+GH-202406,Drain Blockage,Standard,Found relevant keywords: drain,
+GH-202407,Drain Blockage,Standard,Found relevant keywords: drain,
+GH-202410,Other,Standard,No specific keywords mapped.,NEEDS_REVIEW
+GH-202411,Pothole,Standard,Found relevant keywords: pothole,
+GH-202412,Other,Urgent,Found relevant keywords: school,NEEDS_REVIEW
+GH-202417,Other,Standard,"Found relevant keywords: heritage, waste, garbage",NEEDS_REVIEW
+GH-202420,Other,Standard,No specific keywords mapped.,NEEDS_REVIEW
+GH-202422,Pothole,Standard,Found relevant keywords: crater,
+GH-202424,Other,Standard,No specific keywords mapped.,NEEDS_REVIEW
+GH-202428,Waste,Standard,Found relevant keywords: waste,
+GH-202432,Other,Standard,No specific keywords mapped.,NEEDS_REVIEW
+GH-202448,Drain Blockage,Standard,Found relevant keywords: drain,
+GH-202438,Other,Standard,No specific keywords mapped.,NEEDS_REVIEW

uc-0a/results_kolkata.csv

@@ -0,0 +1,16 @@
+complaint_id,category,priority,reason,flag
+KM-202401,Heritage Damage,Standard,Found relevant keywords: heritage,
+KM-202402,Other,Standard,No specific keywords mapped.,NEEDS_REVIEW
+KM-202405,Other,Standard,No specific keywords mapped.,NEEDS_REVIEW
+KM-202409,Other,Standard,No specific keywords mapped.,NEEDS_REVIEW
+KM-202410,Pothole,Standard,Found relevant keywords: pothole,
+KM-202411,Pothole,Standard,Found relevant keywords: pothole,
+KM-202415,Other,Standard,No specific keywords mapped.,NEEDS_REVIEW
+KM-202418,Waste,Standard,Found relevant keywords: waste,
+KM-202421,Other,Urgent,"Found relevant keywords: fell, hospital",NEEDS_REVIEW
+KM-202422,Other,Standard,No specific keywords mapped.,NEEDS_REVIEW
+KM-202426,Heritage Damage,Standard,Found relevant keywords: heritage,
+KM-202430,Other,Standard,No specific keywords mapped.,NEEDS_REVIEW
+KM-202434,Heritage Damage,Standard,Found relevant keywords: heritage,
+KM-202436,Other,Standard,No specific keywords mapped.,NEEDS_REVIEW
+KM-202438,Heritage Damage,Standard,Found relevant keywords: heritage,

uc-0a/results_pune.csv

@@ -0,0 +1,16 @@
+complaint_id,category,priority,reason,flag
+PM-202401,Pothole,Standard,Found relevant keywords: pothole,
+PM-202402,Pothole,Urgent,"Found relevant keywords: pothole, school",
+PM-202406,Other,Standard,No specific keywords mapped.,NEEDS_REVIEW
+PM-202408,Drain Blockage,Standard,Found relevant keywords: drain,
+PM-202410,Streetlight,Standard,Found relevant keywords: dark,
+PM-202411,Streetlight,Urgent,"Found relevant keywords: hazard, streetlight",
+PM-202413,Waste,Standard,Found relevant keywords: garbage,
+PM-202418,Noise,Standard,Found relevant keywords: music,
+PM-202419,Other,Standard,No specific keywords mapped.,NEEDS_REVIEW
+PM-202420,Other,Urgent,Found relevant keywords: injury,NEEDS_REVIEW
+PM-202427,Other,Standard,No specific keywords mapped.,NEEDS_REVIEW
+PM-202428,Other,Standard,No specific keywords mapped.,NEEDS_REVIEW
+PM-202430,Other,Standard,"Found relevant keywords: dark, heritage",NEEDS_REVIEW
+PM-202433,Waste,Standard,Found relevant keywords: waste,
+PM-202446,Other,Urgent,Found relevant keywords: fell,NEEDS_REVIEW

uc-0a/skills.md

@@ -1,15 +1,11 @@
-# skills.md — UC-0A Complaint Classifier
-# INSTRUCTIONS: Same as agents.md — paste README into AI, ask for skills.md YAML
+# Skills
 
-skills:
-  - name: classify_complaint
-    description: "[FILL IN]"
-    input: "[FILL IN]"
-    output: "[FILL IN]"
-    error_handling: "[FILL IN]"
+## `classify_complaint`
+- **Input:** one complaint row (dict with description, location fields)
+- **Output:** dict with category, priority, reason, flag
+- **Error handling:** vague/short descriptions → Other + NEEDS_REVIEW
 
-  - name: batch_classify
-    description: "[FILL IN]"
-    input: "[FILL IN]"
-    output: "[FILL IN]"
-    error_handling: "[FILL IN]"
+## `batch_classify`
+- **Input:** path to test CSV file
+- **Output:** path to results CSV file
+- **Error handling:** malformed rows logged and skipped, processing continues

uc-mcp/agents.md

@@ -1,32 +1,15 @@
-# agents.md — UC-MCP MCP Server
-# INSTRUCTIONS:
-# 1. Open your AI tool
-# 2. Paste the full contents of uc-mcp/README.md
-# 3. Use this prompt:
-#    "Read this UC README. Using the R.I.C.E framework, generate an
-#     agents.md YAML with four fields: role, intent, context, enforcement.
-#     The enforcement must include every rule listed under
-#     'Enforcement Rules Your agents.md Must Include'.
-#     Output only valid YAML."
-# 4. Paste the output below, replacing this placeholder
-# 5. Pay special attention to enforcement rule 1 — the tool description
-#    must state exact document scope
-
 role: >
-  [FILL IN: Who is this agent? What layer of the stack does it operate at?
-   Hint: an MCP server that exposes policy retrieval as a tool]
+  An MCP server operating at the integration layer that exposes semantic search and policy retrieval over CMC policies as a standard MCP tool.
 
 intent: >
-  [FILL IN: What does a correctly implemented MCP server produce?
-   Hint: JSON-RPC compliant responses, scoped tool description, correct refusals]
+  Produces JSON-RPC 2.0 compliant responses, exposes the query_policy_documents tool with a highly specific scoped description, and returns correct refusals for out-of-scope queries.
 
 context: >
-  [FILL IN: What does this server have access to?
-   Hint: RAG server results only — no direct LLM calls, no outside knowledge]
+  Has access to RAG server results only (specifically CMC HR, IT, and Finance policies). Makes no direct LLM calls and has no outside knowledge.
 
 enforcement:
-  - "[FILL IN: Tool description scope rule]"
-  - "[FILL IN: Refusal documentation rule]"
-  - "[FILL IN: inputSchema required field rule]"
-  - "[FILL IN: isError on failure rule]"
-  - "[FILL IN: HTTP 200 for all JSON-RPC responses rule]"
+  - "Tool description must state the exact document scope: CMC HR Leave Policy, IT Acceptable Use Policy, Finance Reimbursement Policy."
+  - "Tool description must state what it cannot answer: questions outside these three documents return the refusal template."
+  - "inputSchema must require `question` as a non-empty string."
+  - "Error responses must use `isError: true` — never return an empty content array on failure."
+  - "The server must return HTTP 200 for all JSON-RPC responses including errors — transport errors use HTTP 4xx/5xx, application errors use JSON-RPC error objects."

uc-mcp/llm_adapter.py

@@ -32,6 +32,12 @@ def call_llm(prompt: str) -> str:
     Call Gemini Flash with the given prompt.
     Returns the text response as a string.
     """
+    try:
+        from dotenv import load_dotenv
+        load_dotenv()
+    except ImportError:
+        pass
+
     api_key = os.environ.get("GEMINI_API_KEY")
     if not api_key:
         return (
@@ -42,7 +48,7 @@ def call_llm(prompt: str) -> str:
     try:
         import google.generativeai as genai
         genai.configure(api_key=api_key)
-        model = genai.GenerativeModel("gemini-1.5-flash")
+        model = genai.GenerativeModel("gemini-2.5-flash")
         response = model.generate_content(prompt)
         return response.text
     except ImportError: