JAR Crime Mapper
Agent User Guide
System Architecture
The dashboard has two cooperating surfaces: map exploration and a planning agent. The agent can run multiple analytics calls and return narratives, charts, tables, and overlays.
- CrimeMapper provides geospatial discovery and map rendering.
- Analysis Agent parses intent, executes tools, then composes a structured report.
- Planner traces expose what was run so outputs are auditable.
System capability scan
Results (0)
| Date | Category | Street | Outcome |
|---|
Interactive Analysis Agent
Ask for charts, comparisons, drilldowns, and suggested next analyses. The agent queries the full crime database through safe server-side analytics tools.
System Status
Loading summary status...
Loading coverage snapshot...
1. Choose a question
Pick a guided question first. You only need to edit the wording if you want something custom.Type or pick a borough
Choose the crime category for guided runs
You are about to run a guided analysis across all boroughs, on all crime.
Filter the guided questions by topic, task, or wording.
What is the big picture?
Quickest path to a plain-English summary of trends, categories, outcomes, and where the pressure sits.Where are the main hotspot areas?
Best first click for broad date ranges; gives the clearest area-level hotspot answer without forcing slow coordinate scans.Which crime type is driving the most pressure?
Best for isolating the leading category and understanding where and how it is concentrated.Which areas need attention most?
Compares boroughs and highlights where pressure, outcomes, or risk look worst.What should teams do first in the active filters?
Best for turning the data into a short operational priority list rather than just describing the problem.What changed recently?
Best for recent movement, changing hotspots, and what shifted versus the prior comparable period.What is driving the hotspot pattern?
Explains the category and outcome mix behind the strongest hotspot areas rather than just ranking them.Which area is carrying the most pressure?
Finds the strongest area or borough in scope and explains the local mix behind it.How are cases being resolved?
Best for understanding outcomes, no-suspect pressure, and whether resolution mix looks healthy.Where is low resolution hurting us most?
Focuses on unresolved outcomes and where resolution quality looks weakest.Which boroughs look similar and which stand out?
Best for finding peer groups, outliers, and unusual pressure patterns across areas.Was there a spike or drop worth explaining?
Looks for the clearest jump or dip in the selected window and explains what caused it.What is getting worse fastest?
Best for finding the categories, places, or outcomes that are deteriorating rather than just high-volume.Are these hotspots persistent or episodic?
Best for testing whether the strongest hotspots keep repeating or are driven by short-lived spikes.What should we watch next?
Forward-looking view of likely trouble spots, uncertainty, and the next things to monitor.Do we have a seasonal or repeating pattern?
Looks for recurring monthly patterns rather than one-off changes.Simple mode shows curated recommended presets across all categories.
3. Review and run
Review the current setup, then run it. You only need the editor below if you want to rewrite the question.Hotspot Map: All Crime
All boroughs | No fixed date window | Category: All crime | Add start and end months | Style: Balanced | Length: Standard | Goal: Automatic | Metrics onSpatial concentration and hotspot confidenceQuestion to run
Press Ctrl+Enter (or Cmd+Enter) to run
Answer style
Most users only need the style below. Open `Customize answer` only if you want more control.Optional override (1-64). Leave blank for auto.
Optional metric-evidence planner budget (0-32).