Daily tiered IP blocklist generated from the ip-radar threat-intelligence engine — 41 sources, consensus verdicts, all evidence recomputed before export.
ip-radar 威胁情报引擎的每日阶梯黑名单导出:41 源聚合、全源共识判定,供防火墙 / fail2ban / 研究直接取用。
Each day's export is published as a GitHub Release with all tiers attached as assets.
Always-latest URL (CDN-backed, stable):
https://github.com/steponeerror/ip-radar-blocklist/releases/latest/download/top_100.txt
Swap top_100 for top_500 / top_1000 / top_5000 / top_10000, or .csv for the enriched format.
| File | Content |
|---|---|
top_100.txt … top_10000.txt |
Pure list: one single IP per line (IPv4 + IPv6 mixed, no CIDR netblocks) |
top_100.csv … top_10000.csv |
Same entries with full context (8 columns) |
Tiers are nested: top_100 ⊂ top_500 ⊂ top_1000 ⊂ top_5000 ⊂ top_10000.
Each tier always exists and holds up to its size ("up to N" — row count is the
honest signal). Updated daily.
ip,asn,country,classes,confidence,source_count,first_seen,last_seen
ip— single IP (v4 or v6); CIDR units are counted in the manifest but excluded from tiersclasses— threat categories (;-joined, e.g.brute-force;scanner)confidence— engine consensus confidence, 0–100source_count— how many independent sources flagged itfirst_seen/last_seen— earliest / latest observation across sources (may be empty)
Entries are ranked by source_count ↓, confidence ↓, recency ↓ — the top of
top_100 is the most heavily corroborated malicious infrastructure.
An IP enters the list only when the engine's fused consensus verdict is
malicious (suspicious-only entries are excluded). No aggregation, no range
expansion. Sources that publish whole netblocks (e.g. Spamhaus DROP, FireHOL
aggregates) are counted in the manifest (cidr_units) but excluded from the
tier files, which carry single IPs only — one IP per line.
# fetch latest
curl -fsSL -o blocklist.txt \
https://github.com/steponeerror/ip-radar-blocklist/releases/latest/download/top_1000.txt
# ipset (iptables)
ipset create ipradar hash:net
ipset restore < blocklist.txt
iptables -I INPUT -m set --match-set ipradar src -j DROPFor fail2ban, point an ipset-backed action at the same URL.
The export is a consensus-derived dataset recomputed from all source evidence and treated as new data — no redistribution restrictions.