🔌 JSON API
Current Azure service tag prefixes, dated diffs and full history as static JSON
Everything the dashboard shows is plain JSON under https://azuretags.egbcorporate.org/data/.
There is no API key, no sign-up and no rate limit. Responses are gzipped, send Access-Control-Allow-Origin: * so browsers can call them directly, and are cached for 10 minutes.
The data is checked against Microsoft's download every day and a new snapshot is recorded whenever Microsoft publishes one, roughly once a week.
This is an independent project, not affiliated with Microsoft. The data itself is Microsoft's ServiceTags_Public.json.
Endpoints
All paths are relative to https://azuretags.egbcorporate.org. Sizes are uncompressed.
| Path | What it contains | Size |
|---|---|---|
/data/current.json | The latest snapshot, in Microsoft's own ServiceTags_Public.json format (changeNumber, cloud, values[]). | 2.9 MB (≈560 KB gzipped) |
/data/changes/manifest.json | Index of every change file, newest first, with its date, change count and Microsoft changeNumber. Start here. | 21 KB |
/data/changes/YYYY-MM-DD-changes.json | What changed since the previous snapshot: the exact prefixes added and removed for each tag. | 1–620 KB |
/data/changes/latest-changes.json | A copy of the newest change file. | varies |
/data/history/YYYY-MM-DD.json | The full snapshot as collected on that date. Same format as current.json. | ≈3 MB each |
/data/summary.json | Totals for the latest snapshot, the list of snapshot dates (available_dates) and a definitions block explaining each count. | 4 KB |
/data/regions.json | Region code to display name, e.g. "australiaeast": "Australia East". | 3 KB |
/data/collection-log.json | Every collection run with its status, plus coverage: which weeks, if any, were missed. | 19 KB |
/tag/<ServiceTag>/ | An HTML page per tag with its prefixes and recent changes, e.g. /tag/Storage.WestEurope/. | HTML |
Tag names are case-sensitive and match Microsoft's: AzureCloud.eastus, Storage.WestEurope.
Get the current prefixes of a tag
curl + jq
curl -s https://azuretags.egbcorporate.org/data/current.json \ | jq -r '.values[] | select(.name == "Storage.WestEurope") | .properties.addressPrefixes[]'
Python
import json, urllib.request
BASE = "https://azuretags.egbcorporate.org/data"
def get(path):
with urllib.request.urlopen(f"{BASE}/{path}") as r:
return json.load(r)
tags = {t["name"]: t for t in get("current.json")["values"]}
for prefix in tags["Storage.WestEurope"]["properties"]["addressPrefixes"]:
print(prefix)
PowerShell
$base = "https://azuretags.egbcorporate.org/data" $tags = (Invoke-RestMethod "$base/current.json").values ($tags | Where-Object name -ceq "Storage.WestEurope").properties.addressPrefixes
See what changed for a tag since a date
Read the manifest, then fetch only the change files you need. Each one already lists the exact prefixes added and removed, so you never have to download and compare 3 MB snapshots yourself.
Python
import json, urllib.request
BASE = "https://azuretags.egbcorporate.org/data"
TAG = "AzureFrontDoor.Backend" # tag names are case-sensitive
SINCE = "2026-09-01" # only changes collected after this date
def get(path):
with urllib.request.urlopen(f"{BASE}/{path}") as r:
return json.load(r)
def diff(change):
"""(added, removed) prefixes for any change type."""
if change["type"] == "service_added":
return change["prefixes"], []
if change["type"] == "service_removed":
return [], change["prefixes"]
return change["added_prefixes"], change["removed_prefixes"]
files = sorted(get("changes/manifest.json")["files"], key=lambda f: f["date"])
for f in files:
if f["baseline"] or f["date"] <= SINCE or f["total_changes"] == 0:
continue
for change in get(f"changes/{f['filename']}")["changes"]:
if change["service"] == TAG:
added, removed = diff(change)
print(f["date"], change["type"], f"+{len(added)} -{len(removed)}")
for p in added:
print(" +", p)
for p in removed:
print(" -", p)
PowerShell
$base = "https://azuretags.egbcorporate.org/data"
$tag = "AzureFrontDoor.Backend" # tag names are case-sensitive
$since = [datetime]"2026-09-01"
$files = (Invoke-RestMethod "$base/changes/manifest.json").files |
Where-Object { -not $_.baseline -and $_.total_changes -gt 0 -and [datetime]$_.date -gt $since } |
Sort-Object { [datetime]$_.date }
foreach ($f in $files) {
$doc = Invoke-RestMethod "$base/changes/$($f.filename)"
foreach ($c in ($doc.changes | Where-Object service -ceq $tag)) {
switch ($c.type) {
'service_added' { $added = @($c.prefixes); $removed = @() }
'service_removed' { $added = @(); $removed = @($c.prefixes) }
default { $added = @($c.added_prefixes); $removed = @($c.removed_prefixes) }
}
"{0} {1} +{2} -{3}" -f $f.filename.Substring(0, 10), $c.type, $added.Count, $removed.Count
$added | ForEach-Object { " + $_" }
$removed | ForEach-Object { " - $_" }
}
}
Find which tags contain an IP address
Works with a single address or a CIDR, IPv4 or IPv6. Expect several hits: a prefix usually belongs to a service tag, its regional tag and the broad AzureCloud tags.
Python
import ipaddress, json, urllib.request
BASE = "https://azuretags.egbcorporate.org/data"
QUERY = ipaddress.ip_network("20.172.90.116/30", strict=False) # an address or a CIDR
with urllib.request.urlopen(f"{BASE}/current.json") as r:
tags = json.load(r)["values"]
for tag in tags:
hits = [p for p in tag["properties"]["addressPrefixes"]
if ipaddress.ip_network(p).overlaps(QUERY)]
if hits:
print(tag["name"], hits)
Or search on the dashboard, which also shows when the range was added.
Get notified of new changes
Run this once a day from cron, a scheduled pipeline or an Azure Function. It remembers the newest change file it has handled and only processes newer ones.
Python
import json, pathlib, urllib.request
BASE = "https://azuretags.egbcorporate.org/data"
STATE = pathlib.Path("last-seen.txt") # newest change file you already processed
def get(path):
with urllib.request.urlopen(f"{BASE}/{path}") as r:
return json.load(r)
last_seen = STATE.read_text().strip() if STATE.exists() else ""
manifest = get("changes/manifest.json")
new = sorted((f for f in manifest["files"] if f["date"] > last_seen and not f["baseline"]),
key=lambda f: f["date"])
for f in new:
doc = get(f"changes/{f['filename']}")
print(f"{f['date']}: {doc['total_changes']} tags changed "
f"(+{doc['comparison']['added_memberships']} / -{doc['comparison']['removed_memberships']} prefixes)")
# ... alert, update firewall rules, open a ticket ...
if new:
STATE.write_text(new[-1]["date"])
On the first run it processes the whole history. To start from today instead, write today's date to last-seen.txt first.
Polling more often is fine. Every file has an ETag, so sending it back with If-None-Match returns an empty 304 Not Modified when nothing changed.
Change file format
Each changes/YYYY-MM-DD-changes.json compares one snapshot with the previous one. All files use "schema_version": 2. New fields may be added, but existing ones will not be renamed or removed without a new schema version.
{
"schema_version": 2,
"date": "2026-11-17",
"baseline": false,
"total_changes": 147,
"changes": [
{
"type": "ip_changes",
"service": "AzureArcInfrastructure",
"region": "",
"system_service": "AzureArcInfrastructure",
"added_prefixes": ["145.133.209.248/30", "145.133.209.252/31"],
"removed_prefixes": [],
"added_count": 88,
"removed_count": 0,
"coverage": {
"ipv4": { "added_addresses": 126, "removed_addresses": 0 },
"ipv6": { "added_slash64s": 0.0, "removed_slash64s": 0.0 }
},
"classification": "expansion"
}
],
"metadata_changes": [],
"metadata": {
"version": "2026.11.16",
"source_url": "https://download.microsoft.com/.../ServiceTags_Public_20261116.json",
"content_sha256": "1e917022…",
"retrieved_at": "2026-11-17T14:30:42+00:00"
},
"comparison": {
"previous_snapshot": "2026-11-15",
"current_snapshot": "2026-11-17",
"previous_change_number": 426,
"current_change_number": 427,
"skipped_change_numbers": [],
"reproducible": true,
"added_memberships": 649,
"removed_memberships": 86,
"balanced": true
}
}
Fields on each change
| Field | Meaning |
|---|---|
type | ip_changes: an existing tag gained or lost prefixes. service_added / service_removed: the whole tag appeared or disappeared. |
service | The tag name, e.g. Storage.WestEurope. |
region, system_service | Copied from the tag. region is "" for global tags. |
added_prefixes, removed_prefixes | On ip_changes: the exact CIDRs that joined or left the tag. |
prefixes, ip_count | On service_added / service_removed: every prefix of the tag, and how many. |
added_count, removed_count | Number of prefixes added and removed, on every type. |
coverage | Address space newly covered and no longer covered: IPv4 in addresses, IPv6 in /64 networks (can be a fraction for prefixes smaller than a /64). |
classification | expansion (only new space), contraction (only lost space), mixed, or representation_only (the CIDRs were rewritten but cover exactly the same addresses). |
File-level fields
| Field | Meaning |
|---|---|
baseline | true only for the first file (2025-10-08), which lists every tag as service_added. Skip it. |
metadata_changes[] | Tags whose region, regionId, platform, systemService or networkFeatures changed, as {"service": …, "fields": {"systemService": {"old": …, "new": …}}}. |
metadata | Which Microsoft file was downloaded, its SHA-256 and when. |
comparison.skipped_change_numbers | Microsoft versions published between the two collections that were never collected. Normally empty. |
comparison.reproducible | true when both snapshots are archived in /data/history/, so the diff can be re-derived. |
comparison.added_memberships, removed_memberships, balanced | Totals across all tags. balanced confirms previous total + added − removed = current total. |
Manifest entries
changes/manifest.json has files[], newest first, each with date, filename, size, change_number, baseline, total_changes, added_memberships, removed_memberships, skipped_change_numbers and reproducible, plus date_range.oldest / date_range.newest.
Good to know
- Counts are memberships. A CIDR listed under three tags counts three times in
total_ip_rangesandadded_memberships.summary.json → definitionsspells this out. - The earliest history is partial. Change files from 2025-10-17 to 2025-11-24 have
reproducible: falseand some have no matching snapshot, because those snapshots were not archived at the time. Their counts are as originally recorded, and some of their entries have nocoverageorclassification, so read those two fields with a default. - Use
summary.json → available_datesto list snapshots. Change file dates and snapshot dates don't line up one-to-one. - Dates are collection dates, not Microsoft's publication date. Microsoft's date is in each change file's
metadata.date_published. - Need Azure China or Government? Only the Public cloud file is tracked.
Found a problem or need another format? Open an issue.