Spring Security
Step 0 — Detect
rg -n "spring-boot|springframework" -g 'pom.xml' -g 'build.gradle*' | head -5
rg --files -g 'application*.properties' -g 'application*.yml' -g '*SecurityConfig*'
Record Spring Boot/Framework version (Spring4Shell check: ../cve-research/vuln-db/entries/), Security config style (WebSecurityConfigurerAdapter vs SecurityFilterChain), template engine.
Step 1 — Configuration files
rg -n "password|secret|include-stacktrace|exposure|cookie" -g 'application*.properties' -g 'application*.yml' -g 'application*.yaml'
spring.datasource.password=.../server.ssl.key-store-password=/ JWT secret hardcoded → Critical in public repos, High otherwisemanagement.endpoints.web.exposure.include=*(or env,heapdump listed) → Critical:/actuator/envleaks every property incl. credentials;/actuator/heapdumpleaks in-memory secretsserver.error.include-stacktrace=always→ Mediumserver.servlet.session.cookie.secure=false/http-only=false→ Medium
Step 2 — Spring Security config
rg -n "csrf\(\)\.disable|permitAll|antMatchers|authorizeHttpRequests|@PreAuthorize|@EnableWebSecurity" src/
http.csrf().disable()with cookie sessions → High (state-changing endpoints).antMatchers("/admin/**").permitAll()→ Critical@PreAuthorize/@PostAuthorizeabsent across controllers with role-sensitive ops → High (verify per-handler)- CORS config:
allowedOrigins("*")+allowCredentials(true)→ Critical (per spec should fail, frameworks reflect)
Step 3 — SQL injection
rg -n "createQuery\([\"'][^\"']*\+|createNativeQuery.*\+|jdbcTemplate\.\w+\([\"'][^\"']*\+" src/
rg -n "\\\$\{" -g '*.xml' src/ # MyBatis mappers
- String-concatenated
createQuery/createNativeQuery/JdbcTemplate.query→ Critical (fix: bound params:name/?) - MyBatis
${...}(string substitution) vs#{...}(prepared) — any${}reachable from request params → Critical
Step 4 — XSS / templates
rg -n "th:utext|<%=|escapeXml=\"false\"|c:out" src/main/resources/templates/ src/ 2>/dev/null
Census, don't sample: disposition every hit. Severity by privilege direction per ../injection-flaws/SKILL.md (XSS table) — unprivileged-authored content rendered unescaped in a staff view = Critical.
th:utext="${user...}"→ direction-triaged (th:textescapes)- JSP scriplets
<%= user %>→ direction-triaged escapeXml="false"on c:out → direction-triaged
Step 5 — Deserialization & SpEL
rg -n "enableDefaultTyping|ObjectInputStream|readObject|parseExpression" src/
- Jackson
enableDefaultTyping()/@JsonTypeInfo(use=Id.CLASS)on external input → Critical (polymorphic gadget chains) ObjectInputStream.readObjecton network data → CriticalspelParser.parseExpression(userInput)→ Critical (SpEL = RCE)- Spring expression in
@Valuefrom constants → safe
Step 6 — Files & commands
rg -n "FileInputStream\(|Files\.(read|copy)|new File\(" src/ | rg -i "request|param" | head
rg -n "Runtime\.getRuntime|ProcessBuilder" src/
- File ops with request params → traversal, High
Runtime.exec(String)single-string form → High (argument injection);exec(String[])/ProcessBuilder list → OK-ish
Reporting
Severity table above; fixes are localized (bound params, th:text, matcher tightening). Cross-reference framework CVEs (Spring4Shell, historic) via ../cve-research/SKILL.md and dependency CVEs via ../dependency-vulns/SKILL.md.