[Thane] [Srikant Ande] - RAG-to-MCP uc-rag Submission

uc-0a/agents.md

@@ -11,17 +11,32 @@
 # 4. Paste the output below
 
 role: >
-  [FILL IN]
+  You are an AI classifier for the City Operations team.
+  You receive hundreds of citizen complaints per week and are responsible for 
+  accurately categorizing them.
 
 intent: >
-  [FILL IN]
+  Read each complaint and output a category, priority, reason, and flag.
+  The output feeds the Director's dashboard every Monday.
+  You must ensure reliable, standardized outputs without hallucinated values.
 
 context: >
-  [FILL IN]
+  You must avoid these five failure modes:
+  1. Taxonomy drift (same complaint type getting different categories).
+  2. Severity blindness (e.g., missing urgent complaints involving children/injuries like "Child fell near school").
+  3. Missing justification (no reason field in output).
+  4. Hallucinated sub-categories (outputting items not in schema, e.g. "Pedestrian Safety Incident").
+  5. False confidence on ambiguity (classifying ambiguous inputs confidently without NEEDS_REVIEW).
+
+  Classification Schema Fields and Allowed Values:
+  - category: Pothole, Flooding, Streetlight, Waste, Noise, Road Damage, Heritage Damage, Heat Hazard, Drain Blockage, Other
+  - priority: Urgent, Standard, Low
+  - reason: One sentence citing specific words from the description
+  - flag: NEEDS_REVIEW or blank
 
 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 of one sentence 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

@@ -10,17 +10,97 @@
 """
 import argparse
 import csv
+import re
+
+SEVERITY_KEYWORDS = {"injury", "child", "school", "hospital", "ambulance", "fire", "hazard", "fell", "collapse"}
+ALLOWED_CATEGORIES = {
+    "Pothole": ["pothole", "crater"],
+    "Flooding": ["flood", "water", "inundated"],
+    "Streetlight": ["light", "lamp", "dark"],
+    "Waste": ["waste", "trash", "garbage", "rubbish"],
+    "Noise": ["noise", "loud", "sound"],
+    "Road Damage": ["road", "crack", "surface"],
+    "Heritage Damage": ["heritage", "monument"],
+    "Heat Hazard": ["heat", "sun"],
+    "Drain Blockage": ["drain", "clog", "blockage", "sewer"],
+}
 
 def classify_complaint(row: dict) -> dict:
     """
     Classify a single complaint row.
     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 for short/vague
+    if len(description.split()) < 5:
+        return {
+            "complaint_id": row.get("complaint_id"),
+            "category": "Other",
+            "priority": "Standard",
+            "reason": "Description is too short or vague.",
+            "flag": "NEEDS_REVIEW"
+        }
+
+    # Check Severity
+    is_urgent = False
+    priority = "Standard"
+    for kw in SEVERITY_KEYWORDS:
+        if kw in description:
+            is_urgent = True
+            priority = "Urgent"
+            break
+
+    # Check Category
+    category = "Other"
+    for cat, aliases in ALLOWED_CATEGORIES.items():
+        if any(alias in description for alias in aliases):
+            category = cat
+            break
+
+    flag = "NEEDS_REVIEW" if category == "Other" else ""
+
+    # Reason
+    sentences = re.split(r'(?<=[.!?])\s+', row.get("description", "").strip())
+    reason_sentence = sentences[0] if sentences else row.get("description")
+    if is_urgent:
+        for s in sentences:
+            if any(kw in s.lower() for kw in SEVERITY_KEYWORDS):
+                reason_sentence = s.strip()
+                break
+    elif category != "Other":
+        for s in sentences:
+            if any(kw in s.lower() for kw in ALLOWED_CATEGORIES[category]):
+                reason_sentence = s.strip()
+                break
+
+    return {
+        "complaint_id": row.get("complaint_id"),
+        "category": category,
+        "priority": priority,
+        "reason": reason_sentence,
+        "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")
+    with open(input_path, 'r', encoding='utf-8') as f:
+        reader = csv.DictReader(f)
+        rows = list(reader)
+
+    results = []
+    fieldnames = ["complaint_id", "category", "priority", "reason", "flag"]
+    for i, row in enumerate(rows):
+        try:
+            res = classify_complaint(row)
+            results.append(res)
+        except Exception as e:
+            print(f"Skipping malformed row {i}: {e}")
+
+    with open(output_path, 'w', encoding='utf-8', newline='') as f:
+        writer = csv.DictWriter(f, fieldnames=fieldnames)
+        writer.writeheader()
+        writer.writerows(results)
 
 if __name__ == "__main__":
     parser = argparse.ArgumentParser(description="UC-0A Complaint Classifier")

uc-0a/results_pune.csv

@@ -0,0 +1,16 @@
+complaint_id,category,priority,reason,flag
+PM-202401,Pothole,Standard,Large pothole 60cm wide causing tyre damage.,
+PM-202402,Pothole,Urgent,School children at risk during morning hours.,
+PM-202406,Flooding,Standard,Underpass flooded knee-deep after 2hrs rain.,
+PM-202408,Flooding,Standard,Bus stand flooded.,
+PM-202410,Streetlight,Standard,Three consecutive streetlights out for 10 days.,
+PM-202411,Streetlight,Urgent,Electrical hazard reported.,
+PM-202413,Waste,Standard,Overflowing garbage bins near vegetable market.,
+PM-202418,Other,Standard,Wedding venue playing music past midnight on weeknights.,NEEDS_REVIEW
+PM-202419,Road Damage,Standard,Road surface cracked and sinking near utility work done 1 month ago.,
+PM-202420,Other,Urgent,Risk of serious injury to cyclists.,NEEDS_REVIEW
+PM-202427,Flooding,Standard,Bridge approach floods in 30mins of rain.,
+PM-202428,Other,Standard,Dead animal not removed for 36 hours.,NEEDS_REVIEW
+PM-202430,Streetlight,Standard,"Heritage street, lights out.",
+PM-202433,Waste,Standard,Bulk waste from apartment renovation dumped on public road.,
+PM-202446,Other,Urgent,Elderly resident fell last week.,NEEDS_REVIEW

uc-0a/skills.md

@@ -3,13 +3,13 @@
 
 skills:
   - name: classify_complaint
-    description: "[FILL IN]"
-    input: "[FILL IN]"
-    output: "[FILL IN]"
-    error_handling: "[FILL IN]"
+    description: "Reads a single citizen complaint and outputs a category, priority, reason, and flag according to the classification schema."
+    input: "One complaint row (dict with description, location fields)"
+    output: "Dictionary with category, priority, reason, flag"
+    error_handling: "If description is vague or short, output category as Other and flag as NEEDS_REVIEW"
 
   - name: batch_classify
-    description: "[FILL IN]"
-    input: "[FILL IN]"
-    output: "[FILL IN]"
-    error_handling: "[FILL IN]"
+    description: "Reads a batch of complaints from an input CSV file, applies classification to each row, and writes the output to a results CSV file."
+    input: "Path to test CSV file"
+    output: "Path to results CSV file"
+    error_handling: "Malformed rows are logged and skipped, processing continues"

uc-rag/agents.md

@@ -1,31 +1,15 @@
-# agents.md — UC-RAG RAG Server
-# INSTRUCTIONS:
-# 1. Open your AI tool
-# 2. Paste the full contents of uc-rag/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, replacing this placeholder
-# 5. Check every enforcement rule against the README before saving
-
 role: >
-  [FILL IN: Who is this agent? What is its operational boundary?
-   Hint: a retrieval-augmented policy assistant for city staff]
+  A retrieval-augmented policy assistant for city staff, designed to answer internal policy questions within strict operational boundaries.
 
 intent: >
-  [FILL IN: What does a correct output look like?
-   Hint: answer + cited chunks + refusal when not covered]
+  To deliver accurate answers supported by cited chunks, or to output a specific refusal template when the policy query is not covered by the retrieved documents.
 
 context: >
-  [FILL IN: What sources may the agent use?
-   Hint: retrieved chunks only — no general knowledge]
+  Answers must be generated using only the retrieved policy document chunks. No general knowledge may be used.
 
 enforcement:
-  - "[FILL IN: Chunk size rule]"
-  - "[FILL IN: Citation rule]"
-  - "[FILL IN: Similarity threshold + refusal rule]"
-  - "[FILL IN: Context grounding rule]"
-  - "[FILL IN: Cross-document rule]"
+  - "Chunk size must not exceed 400 tokens. Never split mid-sentence."
+  - "Every answer must cite the source document name and chunk index."
+  - "If no retrieved chunk scores above similarity threshold 0.6 — output the refusal template: 'This question is not covered in the retrieved policy documents. Retrieved chunks: [list chunk sources]. Please contact the relevant department for guidance.' Never generate an answer from general knowledge."
+  - "Answer must use only information present in the retrieved chunks. Never add context from outside the retrieved set."
+  - "If the query spans two documents — retrieve from each separately. Never merge retrieved chunks from different documents into one answer."