feat(iwooos): 新增 SSH network 事故回讀 gate
This commit is contained in:
Your Name
382
scripts/security/ssh-network-post-incident-readback-plan.py
Normal file
382
scripts/security/ssh-network-post-incident-readback-plan.py
Normal file
@@ -0,0 +1,382 @@
|
||||
#!/usr/bin/env python3
|
||||
"""
|
||||
IwoooS SSH / network / firewall post-incident readback 只讀計畫產生器。
|
||||
|
||||
本工具讀取端口 / 防火牆變更證據驗收 snapshot，建立事故後回讀計畫：
|
||||
誰改、何時改、改前改後、影響哪些服務 / AI provider / public route /
|
||||
monitoring、是否同步相關產品、如何恢復、如何防再發。它不 SSH、不讀 live
|
||||
firewall、不改 port、不做 route smoke、不重啟服務、不打 provider endpoint，
|
||||
也不把事故恢復誤判成 runtime authorization。
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
import argparse
|
||||
import json
|
||||
import subprocess
|
||||
import sys
|
||||
from datetime import datetime, timedelta, timezone
|
||||
from pathlib import Path
|
||||
from typing import Any
|
||||
|
||||
|
||||
TAIPEI = timezone(timedelta(hours=8))
|
||||
|
||||
READBACK_FIELDS = [
|
||||
"readback_candidate_id",
|
||||
"source_change_evidence_candidate_id",
|
||||
"surface_id",
|
||||
"config_kind",
|
||||
"control_tier",
|
||||
"expected_scope",
|
||||
"write_capable_surface",
|
||||
"policy_or_exposure_surface",
|
||||
"change_or_incident_ref",
|
||||
"actor_attribution_ref",
|
||||
"incident_detected_at_ref",
|
||||
"change_window_ref",
|
||||
"affected_port_or_policy_ref",
|
||||
"before_state_ref",
|
||||
"after_state_ref",
|
||||
"service_dependency_ref",
|
||||
"public_route_impact_ref",
|
||||
"ai_provider_impact_ref",
|
||||
"monitoring_alert_impact_ref",
|
||||
"customer_or_product_impact_ref",
|
||||
"operator_notification_ref",
|
||||
"cross_project_sync_ref",
|
||||
"restoration_evidence_ref",
|
||||
"postcheck_readback_ref",
|
||||
"recurrence_guard_ref",
|
||||
"maintenance_window",
|
||||
"rollback_owner",
|
||||
"reviewer_outcome",
|
||||
"followup_owner",
|
||||
"not_approval",
|
||||
]
|
||||
|
||||
REQUIRED_READBACK_FIELDS = [
|
||||
"change_or_incident_ref",
|
||||
"actor_attribution_ref",
|
||||
"incident_detected_at_ref",
|
||||
"change_window_ref",
|
||||
"affected_port_or_policy_ref",
|
||||
"before_state_ref",
|
||||
"after_state_ref",
|
||||
"service_dependency_ref",
|
||||
"public_route_impact_ref",
|
||||
"ai_provider_impact_ref",
|
||||
"monitoring_alert_impact_ref",
|
||||
"customer_or_product_impact_ref",
|
||||
"operator_notification_ref",
|
||||
"cross_project_sync_ref",
|
||||
"restoration_evidence_ref",
|
||||
"postcheck_readback_ref",
|
||||
"recurrence_guard_ref",
|
||||
"maintenance_window",
|
||||
"rollback_owner",
|
||||
"followup_owner",
|
||||
"redacted_evidence_refs",
|
||||
"no_secret_value_attestation",
|
||||
"no_raw_firewall_dump_attestation",
|
||||
"no_false_green_attestation",
|
||||
]
|
||||
|
||||
REVIEWER_CHECKS = [
|
||||
{"check_id": "source_change_evidence_current", "instruction": "來源 change evidence snapshot 必須是目前版本。"},
|
||||
{"check_id": "incident_ref_present", "instruction": "必須有可追溯 incident / change ref。"},
|
||||
{"check_id": "actor_not_anonymous", "instruction": "必須標示 actor role / team，不接受匿名端口關閉。"},
|
||||
{"check_id": "before_after_state_present", "instruction": "必須有變更前與恢復後狀態 ref。"},
|
||||
{"check_id": "port_policy_redacted", "instruction": "端口、policy、host 只收脫敏 ref 或 alias，不保存 raw dump。"},
|
||||
{"check_id": "service_dependency_present", "instruction": "必須列出受影響服務、agent、public route、monitoring 或 deploy path。"},
|
||||
{"check_id": "public_route_impact_present", "instruction": "必須列出 public route / admin route / callback 影響 ref。"},
|
||||
{"check_id": "ai_provider_impact_present", "instruction": "若影響 Ollama / provider health，需列出脫敏 impact ref。"},
|
||||
{"check_id": "monitoring_alert_impact_present", "instruction": "必須列出 alert / SRE / dashboard 影響與 false-green 風險。"},
|
||||
{"check_id": "customer_product_impact_present", "instruction": "需標示產品或使用者影響，不得只寫已恢復。"},
|
||||
{"check_id": "operator_notification_present", "instruction": "必須有受影響 owner / Session / product 的通知 ref。"},
|
||||
{"check_id": "cross_project_sync_present", "instruction": "跨專案同步 ref 必須存在，避免單點修改。"},
|
||||
{"check_id": "restoration_evidence_present", "instruction": "必須有恢復時間與恢復證據 ref。"},
|
||||
{"check_id": "postcheck_independent", "instruction": "post-check 需獨立於原操作人與 UI 卡片。"},
|
||||
{"check_id": "recurrence_guard_present", "instruction": "必須提出防再發 guard、change freeze 或 owner review。"},
|
||||
{"check_id": "emergency_classification_present", "instruction": "緊急破窗需標示分類與事後補件責任。"},
|
||||
{"check_id": "maintenance_window_present", "instruction": "後續任何 port / firewall 操作都需維護窗口。"},
|
||||
{"check_id": "rollback_owner_present", "instruction": "rollback owner 與回復 plan 必須同時存在。"},
|
||||
{"check_id": "no_false_green_route_200", "instruction": "不得只用 route 200 / service up 當成事故已驗收。"},
|
||||
{"check_id": "raw_firewall_dump_absent", "instruction": "不得保存 raw firewall dump、raw iptables、raw nftables 或 raw ACL。"},
|
||||
{"check_id": "secret_or_key_value_absent", "instruction": "不得包含 secret、SSH key、token、cookie、私鑰或 partial secret。"},
|
||||
{"check_id": "hidden_impact_absent", "instruction": "不得隱藏 AI provider、registry、monitoring、deploy 或 product route 影響。"},
|
||||
{"check_id": "counts_transition_safe", "instruction": "只有 reviewer record 能更新 accepted count，且不得同時開 runtime gate。"},
|
||||
{"check_id": "runtime_stays_zero", "instruction": "readback plan 不得觸發任何 SSH、firewall、route smoke、restart 或 production write。"},
|
||||
]
|
||||
|
||||
OUTCOME_LANES = [
|
||||
{"lane_id": "waiting_post_incident_readback", "meaning": "尚未收到事故回讀包；所有 accepted / runtime count 維持 0。"},
|
||||
{"lane_id": "request_actor_supplement", "meaning": "缺 actor / owner / decision 時要求補件。"},
|
||||
{"lane_id": "request_before_after_supplement", "meaning": "缺 before / after 或 restoration evidence 時要求補件。"},
|
||||
{"lane_id": "request_health_impact_supplement", "meaning": "缺 service / AI provider / monitoring / product impact 時要求補件。"},
|
||||
{"lane_id": "quarantine_raw_payload", "meaning": "收到 raw firewall dump、secret 或 key material 時只能隔離。"},
|
||||
{"lane_id": "reject_unattributed_incident", "meaning": "無 actor、無 affected scope、無 rollback 或無 notification 的事故回讀不得驗收。"},
|
||||
{"lane_id": "ready_for_post_incident_review", "meaning": "metadata 合格後，只能進 reviewer review。"},
|
||||
{"lane_id": "incident_readback_only_update", "meaning": "只允許更新只讀 ledger，不得反向視為已批准操作。"},
|
||||
{"lane_id": "recurrence_guard_backfill_required", "meaning": "需補防再發 guard、owner review 與 change freeze。"},
|
||||
{"lane_id": "waiting_runtime_gate", "meaning": "即使 readback accepted，runtime gate 仍需獨立人工批准。"},
|
||||
]
|
||||
|
||||
BLOCKED_ACTIONS = [
|
||||
"ssh_read",
|
||||
"ssh_write",
|
||||
"live_firewall_read",
|
||||
"firewall_change",
|
||||
"port_change",
|
||||
"port_close",
|
||||
"port_open",
|
||||
"network_policy_apply",
|
||||
"nodeport_change",
|
||||
"wireguard_change",
|
||||
"sudo_action",
|
||||
"deploy_ssh_action",
|
||||
"route_smoke",
|
||||
"public_gateway_reload",
|
||||
"nginx_reload",
|
||||
"host_restart",
|
||||
"docker_restart",
|
||||
"systemd_restart",
|
||||
"secret_value_collection",
|
||||
"ssh_key_collection",
|
||||
"raw_firewall_dump_storage",
|
||||
"raw_key_material_storage",
|
||||
"mark_readback_accepted_without_reviewer_record",
|
||||
"mark_incident_resolved_without_postcheck",
|
||||
"hide_cross_project_impact",
|
||||
"treat_route_200_as_all_green",
|
||||
"treat_break_glass_as_approval",
|
||||
"close_management_port_without_owner",
|
||||
"open_runtime_gate",
|
||||
"add_action_button",
|
||||
"production_write",
|
||||
"active_scan",
|
||||
"provider_switch",
|
||||
"prompt_send",
|
||||
]
|
||||
|
||||
|
||||
def git_short_sha(root: Path) -> str:
|
||||
try:
|
||||
result = subprocess.run(
|
||||
["git", "rev-parse", "--short", "HEAD"],
|
||||
cwd=root,
|
||||
check=True,
|
||||
capture_output=True,
|
||||
text=True,
|
||||
)
|
||||
return result.stdout.strip()
|
||||
except Exception:
|
||||
return "unknown"
|
||||
|
||||
|
||||
def load_json(path: Path) -> dict[str, Any]:
|
||||
return json.loads(path.read_text(encoding="utf-8"))
|
||||
|
||||
|
||||
def is_policy_or_exposure(source: dict[str, Any]) -> bool:
|
||||
return source.get("config_kind") in {
|
||||
"k8s_network_policy",
|
||||
"k8s_nodeport_service",
|
||||
"wireguard_runbook",
|
||||
}
|
||||
|
||||
|
||||
def build_candidate(source: dict[str, Any]) -> dict[str, Any]:
|
||||
surface_id = source["surface_id"]
|
||||
return {
|
||||
"readback_candidate_id": f"ssh_network_post_incident_readback:{surface_id}",
|
||||
"status": "waiting_post_incident_readback",
|
||||
"source_change_evidence_candidate_id": source["change_evidence_candidate_id"],
|
||||
"surface_id": surface_id,
|
||||
"config_kind": source["config_kind"],
|
||||
"control_tier": source["control_tier"],
|
||||
"expected_scope": source.get("expected_scope"),
|
||||
"write_capable_surface": source["write_capable_surface"],
|
||||
"policy_or_exposure_surface": is_policy_or_exposure(source),
|
||||
"change_or_incident_ref": None,
|
||||
"actor_attribution_ref": None,
|
||||
"incident_detected_at_ref": None,
|
||||
"change_window_ref": None,
|
||||
"affected_port_or_policy_ref": None,
|
||||
"before_state_ref": None,
|
||||
"after_state_ref": None,
|
||||
"service_dependency_ref": None,
|
||||
"public_route_impact_ref": None,
|
||||
"ai_provider_impact_ref": None,
|
||||
"monitoring_alert_impact_ref": None,
|
||||
"customer_or_product_impact_ref": None,
|
||||
"operator_notification_ref": None,
|
||||
"cross_project_sync_ref": None,
|
||||
"restoration_evidence_ref": None,
|
||||
"postcheck_readback_ref": None,
|
||||
"recurrence_guard_ref": None,
|
||||
"maintenance_window": "pending_post_incident_readback",
|
||||
"rollback_owner": "pending_post_incident_readback",
|
||||
"reviewer_outcome": "waiting_post_incident_readback",
|
||||
"followup_owner": "pending_post_incident_readback",
|
||||
"readback_fields": READBACK_FIELDS,
|
||||
"required_readback_fields": REQUIRED_READBACK_FIELDS,
|
||||
"reviewer_checks": [item["check_id"] for item in REVIEWER_CHECKS],
|
||||
"outcome_lanes": [item["lane_id"] for item in OUTCOME_LANES],
|
||||
"blocked_actions": BLOCKED_ACTIONS,
|
||||
"not_approval": True,
|
||||
"post_incident_readback_received": False,
|
||||
"post_incident_readback_accepted": False,
|
||||
"actor_attribution_accepted": False,
|
||||
"before_after_state_accepted": False,
|
||||
"service_dependency_accepted": False,
|
||||
"public_route_impact_accepted": False,
|
||||
"ai_provider_impact_accepted": False,
|
||||
"monitoring_alert_impact_accepted": False,
|
||||
"operator_notification_accepted": False,
|
||||
"cross_project_sync_accepted": False,
|
||||
"restoration_evidence_accepted": False,
|
||||
"postcheck_readback_accepted": False,
|
||||
"recurrence_guard_accepted": False,
|
||||
"maintenance_window_accepted": False,
|
||||
"rollback_owner_accepted": False,
|
||||
"no_false_green_accepted": False,
|
||||
"ssh_read_authorized": False,
|
||||
"ssh_write_authorized": False,
|
||||
"live_firewall_read_authorized": False,
|
||||
"firewall_change_authorized": False,
|
||||
"port_change_authorized": False,
|
||||
"port_close_authorized": False,
|
||||
"port_open_authorized": False,
|
||||
"network_policy_apply_authorized": False,
|
||||
"nodeport_change_authorized": False,
|
||||
"wireguard_change_authorized": False,
|
||||
"route_smoke_authorized": False,
|
||||
"host_restart_authorized": False,
|
||||
"secret_value_collection_allowed": False,
|
||||
"active_scan_authorized": False,
|
||||
"provider_switch_authorized": False,
|
||||
"prompt_send_authorized": False,
|
||||
"runtime_gate": False,
|
||||
"action_buttons_allowed": False,
|
||||
"production_write_authorized": False,
|
||||
}
|
||||
|
||||
|
||||
def build_report(root: Path, source_report: dict[str, Any], generated_at: str | None) -> dict[str, Any]:
|
||||
report_time = generated_at or datetime.now(TAIPEI).isoformat(timespec="seconds")
|
||||
source_candidates = source_report.get("change_evidence_candidates", [])
|
||||
readback_candidates = [build_candidate(item) for item in source_candidates]
|
||||
write_capable = [item for item in readback_candidates if item["write_capable_surface"]]
|
||||
policy_or_exposure = [item for item in readback_candidates if item["policy_or_exposure_surface"]]
|
||||
|
||||
return {
|
||||
"schema_version": "ssh_network_post_incident_readback_plan_v1",
|
||||
"generated_at": report_time,
|
||||
"git_commit": git_short_sha(root),
|
||||
"status": "post_incident_readback_plan_ready_no_runtime_action",
|
||||
"source_schema_version": source_report.get("schema_version"),
|
||||
"source_status": source_report.get("status"),
|
||||
"source_paths": [
|
||||
"docs/security/PORT-FIREWALL-CHANGE-EVIDENCE-ACCEPTANCE.md",
|
||||
"docs/security/port-firewall-change-evidence-acceptance.snapshot.json",
|
||||
"docs/security/SSH-NETWORK-OWNER-RESPONSE-ACCEPTANCE.md",
|
||||
"docs/security/ssh-network-owner-response-acceptance.snapshot.json",
|
||||
],
|
||||
"summary": {
|
||||
"readback_candidate_count": len(readback_candidates),
|
||||
"write_capable_readback_candidate_count": len(write_capable),
|
||||
"policy_or_exposure_readback_candidate_count": len(policy_or_exposure),
|
||||
"health_impact_review_required_candidate_count": len(readback_candidates),
|
||||
"cross_project_sync_required_candidate_count": len(readback_candidates),
|
||||
"recurrence_guard_required_candidate_count": len(readback_candidates),
|
||||
"readback_field_count": len(READBACK_FIELDS),
|
||||
"required_readback_field_count": len(REQUIRED_READBACK_FIELDS),
|
||||
"reviewer_check_count": len(REVIEWER_CHECKS),
|
||||
"outcome_lane_count": len(OUTCOME_LANES),
|
||||
"blocked_action_count": len(BLOCKED_ACTIONS),
|
||||
"post_incident_readback_received_count": 0,
|
||||
"post_incident_readback_accepted_count": 0,
|
||||
"actor_attribution_accepted_count": 0,
|
||||
"before_after_state_accepted_count": 0,
|
||||
"service_dependency_accepted_count": 0,
|
||||
"public_route_impact_accepted_count": 0,
|
||||
"ai_provider_impact_accepted_count": 0,
|
||||
"monitoring_alert_impact_accepted_count": 0,
|
||||
"operator_notification_accepted_count": 0,
|
||||
"cross_project_sync_accepted_count": 0,
|
||||
"restoration_evidence_accepted_count": 0,
|
||||
"postcheck_readback_accepted_count": 0,
|
||||
"recurrence_guard_accepted_count": 0,
|
||||
"no_false_green_accepted_count": 0,
|
||||
"runtime_gate_count": 0,
|
||||
"action_button_count": 0,
|
||||
"coverage_percent_after_readback_plan": 64,
|
||||
},
|
||||
"required_readback_fields": REQUIRED_READBACK_FIELDS,
|
||||
"reviewer_checks": REVIEWER_CHECKS,
|
||||
"outcome_lanes": OUTCOME_LANES,
|
||||
"blocked_actions": BLOCKED_ACTIONS,
|
||||
"readback_candidates": readback_candidates,
|
||||
"boundaries": {
|
||||
"not_authorization": True,
|
||||
"ssh_read_authorized": False,
|
||||
"ssh_write_authorized": False,
|
||||
"live_firewall_read_authorized": False,
|
||||
"firewall_change_authorized": False,
|
||||
"port_change_authorized": False,
|
||||
"port_close_authorized": False,
|
||||
"port_open_authorized": False,
|
||||
"network_policy_apply_authorized": False,
|
||||
"nodeport_change_authorized": False,
|
||||
"wireguard_change_authorized": False,
|
||||
"route_smoke_authorized": False,
|
||||
"host_restart_authorized": False,
|
||||
"secret_value_collection_allowed": False,
|
||||
"active_scan_authorized": False,
|
||||
"provider_switch_authorized": False,
|
||||
"prompt_send_authorized": False,
|
||||
"runtime_execution_authorized": False,
|
||||
"production_write_authorized": False,
|
||||
"action_buttons_allowed": False,
|
||||
},
|
||||
}
|
||||
|
||||
|
||||
def main() -> int:
|
||||
parser = argparse.ArgumentParser(description=__doc__)
|
||||
parser.add_argument("--root", default=".")
|
||||
parser.add_argument(
|
||||
"--source-change-evidence-report",
|
||||
default="docs/security/port-firewall-change-evidence-acceptance.snapshot.json",
|
||||
)
|
||||
parser.add_argument(
|
||||
"--output",
|
||||
default="docs/security/ssh-network-post-incident-readback-plan.snapshot.json",
|
||||
)
|
||||
parser.add_argument("--generated-at")
|
||||
args = parser.parse_args()
|
||||
|
||||
root = Path(args.root).resolve()
|
||||
source_report = load_json(root / args.source_change_evidence_report)
|
||||
report = build_report(root, source_report, args.generated_at)
|
||||
payload = json.dumps(report, ensure_ascii=False, indent=2, sort_keys=True)
|
||||
|
||||
output_path = root / args.output
|
||||
output_path.parent.mkdir(parents=True, exist_ok=True)
|
||||
output_path.write_text(payload + "\n", encoding="utf-8")
|
||||
|
||||
summary = report["summary"]
|
||||
print(
|
||||
"SSH_NETWORK_POST_INCIDENT_READBACK_PLAN_OK "
|
||||
f"candidates={summary['readback_candidate_count']} "
|
||||
f"checks={summary['reviewer_check_count']} "
|
||||
f"lanes={summary['outcome_lane_count']} "
|
||||
f"accepted={summary['post_incident_readback_accepted_count']} "
|
||||
f"runtime_gate={summary['runtime_gate_count']}"
|
||||
)
|
||||
return 0
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
sys.exit(main())
|
||||
Reference in New Issue
Block a user