加密流量元数据与行为分析

EncGuard Pro 加密通信恶意行为检测平台

检测模式：规则检测 + 分类模型 + 异常检测 + 攻击链融合 模型状态：已启用

Alert Detail

告警详情

runtime-615c6a39-715b-40ee-aace-48654fda9cdc

模型预测与异常检测已拆分为 model_prediction 与 anomaly_result 两个解释区。

风险评分 95 critical
Basic Info

基础信息

event_id
runtime-615c6a39-715b-40ee-aace-48654fda9cdc
flow_id
safe-lab-20260704092453-e82171-028-sftp_bulk_upload_like
src_ip
10.22.6.216
dst_ip
203.0.113.202
protocol
sftp
timestamp
2026-07-04T09:24:55.493068+00:00
风险等级
critical
检测类型
data_exfiltration
detectors
rule_engine, anomaly_detector
source
mixed_scenario_lab
chain_related
False
Risk Score

风险评分

final_risk_score
95
severity
critical
security_alert
True
Risk Formation

最终风险形成

rule_score
95
supervised_risk_score
0
calibrated_risk_score
17
anomaly_risk_score
95
final_risk_score
95
fusion_policy
max(rule_score, calibrated_risk_score, anomaly_risk_score)
gate_formula
risk=(1-g)*supervised_risk+g*anomaly_risk
Weak Scenario Support

弱覆盖场景补强证据

scenarios
data_exfiltration
support_type
rule_anomaly_gate_attack_chain_supported
model_claim
not_strong_supervised_label
explanation
This scenario is supported by rules, anomaly detection, uncertainty gate, and chain evidence rather than claimed as a strongly supervised classifier label.
Model Prediction

模型预测 model_prediction

predicted_label
benign
malicious_probability
0.0047
model_score
0
calibrated_risk_score
17
uncertainty_gate
0.183467
probability_entropy
0.037853
top2_margin
0.989567
js_divergence
0.001524
unknown_decision
False
model_supported_label
True

该场景主要由规则、攻击链或异常检测补强，不应表述为当前模型可靠覆盖。

Anomaly Detection

异常检测 anomaly_result

anomaly_score
97.721204
anomaly_risk_score
95
is_anomaly
True
Rule Hits

规则 evidence

data_exfiltration scenario_rule_engine · risk_score=95
Evidence

推荐处置 recommendation

  • uploaded_file=mixed_scenario_features.csv
  • input_label=data_exfiltration
  • uploaded feature row evaluated by EncGuard scenario rule engine
  • anomaly_score=97.721204
  • anomaly_risk_score=90.0
  • uncertainty_gate=0.183467
  • safe_lab_batch=safe-lab-20260704092453-e82171
  • scenario=sftp_bulk_upload_like
  • ground_truth_label=data_exfiltration
  • safe synthetic metadata only; no payload and no real attack execution

recommendation：Review uploaded feature context and validate whether this communication is authorized.

Attack Chain

攻击链关联 attack_chain

暂无攻击链关联。

Safety Boundary

安全边界说明

  • EncGuard 不解密 SSH/TLS/HTTPS/SFTP/SCP payload。
  • 当前检测依据为连接元数据、流量统计、规则 evidence、模型预测、异常检测和攻击链聚合。
  • post_bruteforce_action、encrypted_tunnel、data_exfiltration 主要由规则、攻击链或异常检测补强。
Raw Feature

原始特征详情

flow_id
safe-lab-20260704092453-e82171-028-sftp_bulk_upload_like
src_ip
10.22.6.216
dst_ip
203.0.113.202
src_port
63626
dst_port
22
protocol
sftp
start_time
2026-07-04T09:49:22Z
duration
885.2763
total_packets
492576
total_bytes
1012933842
fwd_packets
447605
bwd_packets
44971
fwd_bytes
1006103036
bwd_bytes
6830806
upload_download_ratio
147.289066
packets_per_second
556.409314
bytes_per_second
1144200.740901
packet_size_mean
2056.401128
packet_size_std
2318.844897
iat_mean
0.010708
iat_std
0.002342
same_src_conn_count_1min
2
same_src_conn_count_5min
4
short_connection_count_5min
0
long_session_flag
False
large_upload_flag
True
unknown_destination_flag
True
non_working_hour_flag
True
packet_direction_sequence_summary
client_to_server_dominant
packet_size_sequence_summary
large_forward_bulk_transfer_like
burst_count
94
interactive_gap_mean
0.010708
interactive_gap_std
0.002342
post_login_long_session_flag
False
command_like_burst_flag
False
failed_login_then_long_session_flag
False
scenario
sftp_bulk_upload_like
label
data_exfiltration
Related Events

相关事件

同源 IP 相关事件

暂无同源相关事件。

同目的 IP 相关事件

暂无同目的相关事件。