php-logic-audit
Compare original and translation side by side
🇺🇸
Original
English🇨🇳
Translation
ChinesePHP 业务逻辑漏洞审计(php-logic-audit)
PHP Business Logic Vulnerability Audit (php-logic-audit)
分析 PHP 项目中“业务流程正确性”相关缺陷。重点覆盖:
- Mass Assignment(批量赋值导致可越权字段被写入)
- 流程绕过(前置状态缺失或校验可被跳过)
- 状态机缺陷(状态更新不完整/可回滚/不一致)
- 竞态条件(TOCTOU、并发导致多次发放、重复扣款等)
- 越权的业务层校验缺失(与 IDOR 联动,但更关注业务归属一致性)
- 付款/退款/异步任务顺序漏洞(异步回调可被伪造或重复处理)
Analyzes defects related to "business process correctness" in PHP projects. Key coverage includes:
- Mass Assignment (overprivileged fields can be written due to batch assignment)
- Process Bypass (missing preconditions or skippable validations)
- State Machine Defects (incomplete/rollbackable/inconsistent state updates)
- Race Conditions (TOCTOU, multiple distributions or repeated deductions caused by concurrency, etc.)
- Missing Business-layer Validation for Privilege Escalation (linked with IDOR, but focuses on business ownership consistency)
- Payment/Refund/Asynchronous Task Sequencing Vulnerabilities (asynchronous callbacks can be forged or repeatedly processed)
分级与编号
Severity Rating and Vulnerability ID
- 详见:
shared/SEVERITY_RATING.md - 漏洞编号:
{C/H/M/L}-LOGIC-{序号}
- See details:
shared/SEVERITY_RATING.md - Vulnerability ID format:
{C/H/M/L}-LOGIC-{serial number}
必检证据(强制)
Mandatory Evidence Collection
每条逻辑漏洞必须提供:
- 触发路由与方法(真实路由/HTTP 方法)
- 状态前置条件(攻击者需要满足哪些条件:登录态/角色/资源状态)
- 缺陷点位置(业务服务/控制器/回调 handler/队列消费逻辑)
- 状态变化链(从输入 -> 校验 -> 写库/调用外部服务 -> 状态更新)
- 与并发/时序相关证据(如果有:事务边界/锁/唯一约束/幂等处理)
Each logic vulnerability must provide:
- Trigger Route and Method (real route/HTTP method)
- Preconditions (conditions attackers need to meet: login status/role/resource status)
- Defect Location (business service/controller/callback handler/queue consumption logic)
- State Change Chain (from input -> validation -> database write/external service call -> state update)
- Concurrency/Timing-related Evidence (if applicable: transaction boundaries/locks/unique constraints/idempotency handling)
Mass Assignment 审计(必做)
Mass Assignment Audit (Mandatory)
必须检测:
- 是否存在把 直接写入模型或数据库的行为
$_POST/Request->all()/request()->all() - 是否存在允许更新敏感字段的白名单缺失(如
role/is_admin/status/price/owner_id - ORM/框架的填充保护是否启用(例如 Laravel )
$fillable/$guarded
Must detect:
- Whether there are behaviors directly writing to models or databases
$_POST/Request->all()/request()->all() - Whether there is a missing whitelist that allows updating sensitive fields (e.g., )
role/is_admin/status/price/owner_id - Whether ORM/framework fill protection is enabled (e.g., Laravel )
$fillable/$guarded
竞态与幂等审计(必做)
Race Condition and Idempotency Audit (Mandatory)
必须检测:
- 同一操作是否在并发下被执行多次(扣款、发货、升级、领取)
- 是否依赖“先查再写”(TOCTOU)且缺少事务/行锁
- 是否有幂等键与去重(如 payment_id/nonce)
Must detect:
- Whether the same operation can be executed multiple times under concurrency (deduction, delivery, upgrade, redemption)
- Whether it relies on "check-then-write" (TOCTOU) without transactions/row locks
- Whether there are idempotency keys and deduplication mechanisms (e.g., payment_id/nonce)
流程绕过审计(必做)
Process Bypass Audit (Mandatory)
必须检测:
- 是否只在前置页面校验但后端缺少校验
- 是否依赖前端隐藏字段/按钮来限制流程
- 回调接口是否鉴权与签名可靠、是否可被重复调用
Must detect:
- Whether validation is only performed on the front-end page but missing on the back-end
- Whether process restrictions rely on front-end hidden fields/buttons
- Whether callback interfaces have reliable authentication and signatures, and whether they can be repeatedly called
类型比较绕过与弱类型漏洞(必做)
Type Comparison Bypass and Weak Type Vulnerabilities (Mandatory)
PHP 中的弱类型/宽松比较经常导致认证/授权/业务判断绕过,必须审计:
- 是否存在使用 /
==(非严格比较)进行鉴权判断、状态判断或权限校验!= - 是否存在 未设置严格模式(第三参),导致类型转换绕过
in_array($x, $arr) - 是否存在 、
array_search等结果参与条件判断时未正确处理(例如strpos返回 0 被当成 false,或与字符串/数字混用)strpos - 是否存在 用于“认证校验”且直接用
md5/sha1比较摘要(必须识别是否使用==)hash_equals - 是否存在 token/JWT/验证码比较逻辑使用了可被时序推断或类型胶水绕过的方式(比较函数/条件表达式需证据化)
每条弱类型漏洞必须给出:
- 位置证据(比较表达式/条件语句/相关函数调用位置)
- Source(用户输入如何进入该比较)
- Sink(错误比较导致的分支结果:放行/绕过/错误状态)
- PoC(请求序列或 payload,至少包含“对比值如何构造触发绕过”)
Weak type/loose comparison in PHP often leads to authentication/authorization/business judgment bypasses, so the following must be audited:
- Whether /
==(non-strict comparison) is used for authentication judgment, state judgment or permission validation!= - Whether is used without strict mode (third parameter), leading to type conversion bypass
in_array($x, $arr) - Whether results of ,
array_searchetc. are improperly handled when participating in condition judgments (e.g.,strposreturning 0 is treated as false, or mixed with strings/numbers)strpos - Whether is used for "authentication validation" and digests are directly compared with
md5/sha1(must identify whether==is used)hash_equals - Whether token/JWT/verification code comparison logic uses methods that can be bypassed via timing inference or type juggling (comparison functions/conditional expressions need to be evidenced)
Each weak type vulnerability must provide:
- Location Evidence (comparison expression/conditional statement/relevant function call location)
- Source (how user input enters the comparison)
- Sink (branch result caused by incorrect comparison: pass/bypass/error state)
- PoC (request sequence or payload, including at least "how to construct the comparison value to trigger bypass")
认证/账户工作流逻辑(必做)
Authentication/Account Workflow Logic (Mandatory)
必须审计与“状态迁移”相关的业务流程安全:
- 密码重置 token 是否单次使用、是否有明确过期时间、是否与用户绑定
- OTP/验证码是否存在重放(同一 token/nonce 是否能重复使用)
- 登录失败次数/账户锁定/速率限制是否实现且发生在后端
- 账户枚举:错误信息/响应码/耗时差异是否泄露是否存在用户
- 支付/退款回调是否幂等(同一 transaction 是否重复入账/重复发放)
PoC 必须给出至少两步请求序列:先建立前置状态,再触发状态迁移动作,并标注观测点(数据库字段/响应码/重定向/回调执行次数)。
Must audit business process security related to "state migration":
- Whether password reset tokens are single-use, have clear expiration times, and are bound to users
- Whether OTP/verification codes can be replayed (whether the same token/nonce can be reused)
- Whether login failure count/account lockout/rate limiting are implemented on the back-end
- Account Enumeration: whether error messages/response codes/latency differences leak the existence of users
- Whether payment/refund callbacks are idempotent (whether the same transaction can be repeatedly recorded/disbursed)
PoCs must provide at least a two-step request sequence: first establish preconditions, then trigger state migration actions, and mark observation points (database fields/response codes/redirects/callback execution counts).
PoC(强制)
PoC (Mandatory)
逻辑漏洞 PoC 必须给出“请求序列”:
- 至少两步:触发前置条件 -> 触发漏洞动作(并标注真实路由)
- 若是竞态漏洞:给出并发请求方式与观测点(例如两个 curl 同时发送的步骤)
PoCs for logic vulnerabilities must provide a "request sequence":
- At least two steps: trigger preconditions -> trigger vulnerability action (mark real routes)
- For race condition vulnerabilities: provide concurrent request methods and observation points (e.g., steps for sending two curl requests simultaneously)
报告输出
Report Output
{output_path}/vuln_audit/logic_{timestamp}.md{output_path}/vuln_audit/logic_{timestamp}.md条目模板(强制)
Entry Template (Mandatory)
- 位置证据
- 缺陷描述(必须基于代码语义,不可只用一般性话术)
- 状态变化链
- 可利用前置条件
- 请求序列 PoC
- 修复建议(包括:白名单/事务/锁/幂等/后端二次校验)
- Location Evidence
- Defect Description (must be based on code semantics, general statements are not allowed)
- State Change Chain
- Exploitable Preconditions
- Request Sequence PoC
- Remediation Suggestions (including: whitelist/transaction/lock/idempotency/back-end revalidation)