| name | forest-monitoring |
| description | Query forest health data for any location using Global Forest Watch. Use when users ask about deforestation, fires, tree cover, or forest health near them or near their observations. |
Forest Monitoring
When to Use
- User asks "how is the forest doing near me?" or similar
- User asks about deforestation, fires, tree cover loss
- After publishing an observation, to add forest context ("your toucan was spotted in an area that lost 35% of its forest")
- User shares a location and asks about environmental threats
- User asks about their municipality, district, or region's forest health
How to Use
- You need GPS coordinates. If the user shared a Telegram location, use those. If they gave a place name, use geocode_location first.
- Call forest_report with latitude and longitude. The tool automatically identifies the municipality, state, and country for the location.
- Present the results conversationally — do NOT dump raw numbers
Presenting Results
- Lead with the area name: Forest Report for Altamira, Pará, Brazil
- Lead with the most striking finding (e.g., "Your area has lost 35% of its forest since 2000")
- The report covers the entire municipality — mention this so users understand the scope
- Use emoji sparingly: 🌳 for tree cover, 🔥 for fires, 🛰️ for deforestation alerts
- Round numbers for readability (314.67 ha → "about 315 hectares")
- Convert hectares to something relatable when useful ("that is about 440 football fields")
- If there are active fires or recent deforestation, highlight that urgently
- If the area is healthy (low loss, no alerts), celebrate it
- If the user asks about a broader area (state, country), mention that the data is available at those levels too
Attribution
Always include "Data: Global Forest Watch" at the end of forest reports.
Don't
- Don't show raw JSON to users
- Don't present all data points — pick the 2-3 most meaningful
- Don't lecture about deforestation — let the numbers speak
- Don't call forest_report unless the user asked about forest/environment topics or you're adding context to an observation