Web Agent

Chat with the AI agent directly inside AnomalyArmor

The Web Agent is an AI assistant built into AnomalyArmor’s Intelligence page. It has access to 30+ tools across 9 domains and can chain them together to investigate issues, set up monitoring, manage alerts, and answer questions about your data.

Getting Started

1. Navigate to any asset’s Intelligence tab
2. Click Ask Agent in the top right
3. Type a question or request in plain English

The agent opens as a full-screen overlay with your Intelligence page still accessible behind it.

What Makes It Powerful

The Web Agent isn’t just a chatbot. When you ask a question, it:

1. Reads your monitoring context: Schema drift history, freshness status, data quality metrics, tags, lineage, and alert rules for the selected asset
2. Chains tools automatically: A single question can trigger multiple tool calls across domains
3. Shows its work: Every tool call is visible with parameters and results, so you can verify what the agent is doing

Example: Diagnosing a Problem

You: Why is the orders dashboard broken?

Agent: Let me investigate...

[Tool: list_schema_changes > gold.fact_orders]
  Found: discount_pct column removed, Tuesday 4:12 PM

[Tool: check_freshness > gold.fact_orders]
  Status: FRESH (last update 12 minutes ago)

[Tool: get_lineage > gold.fact_orders]
  Upstream: staging.orders_raw
  Downstream: analytics.order_summary, reporting.daily_revenue

[Tool: list_schema_changes > staging.orders_raw]
  Found: discount_pct column removed, Tuesday 3:47 PM

The dashboard is broken because the discount_pct column was removed
from gold.fact_orders on Tuesday at 4:12 PM.

Root cause: The column was first removed from staging.orders_raw at
3:47 PM, then the ETL propagated the removal downstream.

2 views reference this column:
- analytics.order_summary
- reporting.daily_revenue

Freshness is normal, so the table is still updating. The issue is
the missing column, not stale data.

The agent called 4 tools across 3 domains (schema drift, freshness, lineage) to build a complete diagnosis from a single question.

Capabilities by Domain

Discovery

Browse your data assets without leaving the chat.

You: What tables do I have in the gold schema?

Agent: Found 12 tables in warehouse.gold:

       Fact tables:
       - gold.fact_orders (3.2M rows, fresh)
       - gold.fact_payments (1.8M rows, fresh)
       - gold.fact_refunds (245K rows, fresh)
       - gold.fact_subscriptions (450K rows, fresh)

       Dimension tables:
       - gold.dim_customer (245K rows, fresh)
       - gold.dim_product (18K rows, fresh)
       - gold.dim_geography (2.1K rows, fresh)
       ... and 5 more

Monitoring Setup

Set up freshness schedules, schema monitoring, and data quality metrics in one conversation.

You: Set up monitoring for the gold fact tables. I want hourly
     freshness checks and null_percentage metrics on any amount columns.

Agent: Done. Here's what I set up:

       Freshness schedules (4 tables):
       - gold.fact_orders: hourly, 2h SLA
       - gold.fact_payments: hourly, 2h SLA
       - gold.fact_refunds: hourly, 2h SLA
       - gold.fact_subscriptions: hourly, 2h SLA

       Data quality metrics (6 columns):
       - gold.fact_orders.order_total: null_percentage
       - gold.fact_orders.discount_amount: null_percentage
       - gold.fact_payments.amount: null_percentage
       - gold.fact_payments.fee_amount: null_percentage
       - gold.fact_refunds.refund_amount: null_percentage
       - gold.fact_subscriptions.subscription_value: null_percentage

       Want me to add alert rules for these metrics?

Alert Management

View, acknowledge, resolve, or dismiss alerts. The agent shows context from monitoring data alongside each alert.

You: What alerts fired today? Dismiss any info-level ones.

Agent: Found 5 alerts today.

       Dismissed 2 info-level alerts:
       - Schema baseline captured for gold.dim_geography
       - Freshness schedule activated for staging.email_events

       Remaining 3 alerts:
       1. [CRITICAL] Freshness SLA breach
          gold.fact_orders - 6 hours overdue (SLA: 2h)
          Upstream staging.orders_raw is also stale.

       2. [WARNING] Data quality anomaly
          gold.dim_customer.email - null rate 2.1% > 12.4%
          Gradual increase over 5 days, not a sudden spike.

       3. [WARNING] Schema drift detected
          gold.fact_orders - discount_pct column removed
          2 downstream views affected.

Data Quality

Create metrics and validity rules. The agent asks for confirmation before creating anything.

You: Add null checks on all email columns across my tables

Agent: I found email columns in 4 tables:

       - gold.dim_customer.email (current null rate: 12.4%)
       - silver.silver_customers.email_address (current: 0.4%)
       - raw.user_signups.email (current: 0.0%)
       - staging.email_events.recipient_email (current: 1.2%)

       I'll create null_percentage metrics for each. Proceed? [Yes/No]

You: Yes

Agent: Created 4 null_percentage metrics.

       Note: gold.dim_customer.email already has an elevated null rate
       (12.4%). You may want to investigate before setting alert
       thresholds, otherwise you'll get an immediate alert.

Schema Drift

Review changes, trace their origin, and understand downstream impact.

You: What schema changes happened this week? Show me the impact.

Agent: 3 schema changes detected this week:

       1. gold.fact_orders.discount_pct REMOVED (Tuesday 4:12 PM)
          Impact: 2 downstream views reference this column
          Origin: Removed from staging.orders_raw at 3:47 PM

       2. gold.dim_customer.loyalty_tier ADDED (Wednesday 9:30 AM)
          Type: VARCHAR(50)
          Impact: No downstream references yet (new column)

       3. silver.silver_products.weight REMOVED (Monday 2:15 PM)
          Impact: 1 downstream table (gold.dim_product) joins on this
          Origin: Source table raw.products no longer has this column

Tags and Classification

Apply and manage tags for organization and compliance.

You: Find all tables with email columns and tag them as PII

Agent: Found email columns in 4 tables. Applied PII tag:

       - gold.dim_customer (column: email)
       - silver.silver_customers (column: email_address)
       - raw.user_signups (column: email)
       - staging.email_events (column: recipient_email)

       These tables now appear when you ask "which tables have PII?"
       and will be flagged in compliance reports.

Sessions

Conversations are organized into sessions. Each session maintains full context, so follow-up questions work naturally.

• New session: Click + in the session sidebar
• Resume session: Click a previous session to continue where you left off
• Pin session: Pin important sessions to the top of the list
• Session history: Sidebar shows all past conversations with timestamps and message counts

Tool Visibility

As the agent works, you see each tool call in real time:

• Which tool was called (e.g., list_schema_changes, check_freshness)
• What parameters were passed
• The result returned

For actions that modify data (creating metrics, applying tags, acknowledging alerts), the agent asks for confirmation before proceeding.

Tips

• Be specific about scope: “Check freshness for the 3 gold fact tables” works better than “check everything”
• Chain requests: “Find tables with spiking null rates, then show me their schema changes” works in one message
• Ask diagnostic questions: “Why is X broken?” triggers cross-domain investigation, which is where the agent is strongest
• Let it discover first: Say “find tables with customer data” instead of typing exact table paths
• Review before acting: For bulk operations, ask “what tables match?” before “set up monitoring for all of them”

Available Tools

Domain	Count	Capabilities
Discovery	7	Browse assets, catalogs, schemas, tables, columns
Freshness	4	Check freshness, find stale tables, create schedules, get recommendations
Schema	3	List changes, view details, enable monitoring
Data Quality	3	Create metrics, list existing, view health summary
Validity	3	Create rules, list existing, view results
Alerts	4	List alerts, view details, list rules, create rules
Alert Actions	5	Acknowledge, resolve, dismiss, snooze alerts
Tags	2	List and apply tags
Health	1	Overall health summary

Next Steps

Intelligence Overview

How monitoring context powers the agent’s answers

MCP Server

Use the same tools from your IDE

Status Skill

Health monitoring details

Alerts Skill

Alert management details