import type { RadarMCP, UserDetails } from './radar.app'
export interface Env {
OAUTH_KV: KVNamespace
MCP_COOKIE_ENCRYPTION_KEY: string
ENVIRONMENT: 'development' | 'staging' | 'production'
MCP_SERVER_NAME: string
MCP_SERVER_VERSION: string
CLOUDFLARE_CLIENT_ID: string
CLOUDFLARE_CLIENT_SECRET: string
MCP_OBJECT: DurableObjectNamespace<RadarMCP>
USER_DETAILS: DurableObjectNamespace<UserDetails>
MCP_METRICS: AnalyticsEngineDataset
DEV_DISABLE_OAUTH: string
DEV_CLOUDFLARE_API_TOKEN: string
DEV_CLOUDFLARE_EMAIL: string
}
export const BASE_INSTRUCTIONS = /* markdown */ `
# Cloudflare Radar MCP Server
This server provides tools powered by the Cloudflare Radar API for global Internet insights.
## Authentication
- **URL Scanner** requires an active account (use \`set_active_account\`)
- All other Radar data tools work without account selection
## Tool Categories
- **Entities**: Look up ASN, IP, and location details (\`list_autonomous_systems\`, \`get_as_details\`, \`get_ip_details\`)
- **Traffic**: HTTP and DNS trends (\`get_http_data\`, \`get_dns_queries_data\`)
- **Attacks**: Layer 3/7 DDoS attack trends (\`get_l3_attack_data\`, \`get_l7_attack_data\`)
- **Email**: Routing and security trends (\`get_email_routing_data\`, \`get_email_security_data\`)
- **Quality**: Internet speed and quality metrics (\`get_internet_quality_data\`, \`get_internet_speed_data\`)
- **Rankings**: Top domains and services (\`get_domains_ranking\`, \`get_internet_services_ranking\`)
- **AI**: AI bot traffic and Workers AI usage (\`get_ai_data\`)
- **BGP**: Route hijacks, leaks, and stats (\`get_bgp_hijacks\`, \`get_bgp_leaks\`, \`get_bgp_route_stats\`)
- **Bots**: Bot traffic by category, operator, kind (\`get_bots_data\`)
- **Certificate Transparency**: SSL/TLS certificate issuance trends (\`get_certificate_transparency_data\`)
- **NetFlows**: Network traffic patterns with ADM1 filtering (\`get_netflows_data\`)
- **Cloud Observatory**: Cloud provider performance - AWS, GCP, Azure, OCI (\`list_origins\`, \`get_origin_details\`, \`get_origins_data\`)
- **URL Scanner**: Scan and analyze URLs for security threats (\`search_url_scans\`, \`create_url_scan\`, \`get_url_scan\`, \`get_url_scan_screenshot\`, \`get_url_scan_har\`)
## Making Comparisons
Many tools support **array-based filters** for comparisons. Filter arrays (location, asn, continent, geoId) map **positionally** to dateRange arrays - each array index corresponds to a distinct data series.
### Cross-Location Comparison (Same Time Period)
Compare HTTP traffic between Portugal and Spain over the last 7 days:
- \`dateRange: ["7d", "7d"]\`
- \`location: ["PT", "ES"]\`
Result: \`summary_0\` = Portugal (7d), \`summary_1\` = Spain (7d)
### Time Period Comparison (Same Location)
Compare Portugal bandwidth this week vs last week:
- \`dateRange: ["7d", "7dcontrol"]\`
- \`location: ["PT", "PT"]\` *(repeat the location for each dateRange)*
Result: \`summary_0\` = PT (current week), \`summary_1\` = PT (previous week)
### IMPORTANT: Positional Mapping Behavior
Filter arrays must match the length of dateRange for consistent filtering. If you provide fewer filter values than dateRange values, unmatched periods default to worldwide/unfiltered data.
**Correct** (comparing Portugal across time periods):
- \`dateRange: ["7d", "7dcontrol"], location: ["PT", "PT"]\`
**Incorrect** (control period will be worldwide, not Portugal):
- \`dateRange: ["7d", "7dcontrol"], location: ["PT"]\` *(missing second location)*
This positional mapping applies to all filter arrays: **location**, **asn**, **continent**, and **geoId**.
## Geographic Options
- **location**: Filter by country (alpha-2 codes like "US", "PT")
- **continent**: Filter by continent (alpha-2 codes like "EU", "NA")
- **asn**: Filter by Autonomous System Number (e.g., "13335" for Cloudflare, "15169" for Google)
- **geoId**: Filter by ADM1 region (GeoNames IDs for states/provinces) - available for HTTP and NetFlows
## Dimension Types
Many tools use a \`dimension\` parameter to control the response format:
### timeseries
Returns data points over time with timestamps and values. Best for line charts showing trends.
\`\`\`
{ timestamps: ["2026-01-01T12:00:00Z", ...], values: ["0.735", "0.812", ...] }
\`\`\`
### summary
Returns aggregated percentages or values grouped by a category. Best for pie/bar charts.
\`\`\`
{ Chrome: "31.44%", ChromeMobile: "27.89%", Safari: "12.27%", ... }
\`\`\`
### timeseriesGroups
Combines both: time series data broken down by category. Best for stacked area/line charts.
\`\`\`
{ timestamps: [...], Chrome: ["28.15", "27.35", ...], Firefox: ["31.01", "31.23", ...] }
\`\`\`
Use \`summary/*\` dimensions (e.g., \`summary/browser\`) for snapshots and \`timeseriesGroups/*\` (e.g., \`timeseriesGroups/browser\`) for trends over time.
## Visualizations
Generate charts when appropriate:
- **Line charts**: Timeseries data
- **Bar charts**: Rankings, summaries
- **Pie charts**: Distributions
- **Choropleth maps**: Geographic data
- **Stacked area charts**: Grouped timeseries
`
