security-scan

name	security-scan
description	Security scanning domain knowledge — OWASP Top 10, CWE mappings, IaC hardening, CI/CD pipeline security, supply chain controls, and SARIF output format

Security Scan Skill

Domain knowledge for security scanning agents. Agents load this skill to understand vulnerability categories, severity classification, compliance mappings, and SARIF output requirements.

OWASP Top 10 Vulnerability Categories

The following table maps OWASP Top 10 (2021) categories to their primary CWE identifiers for automated detection and SARIF enrichment.

#	OWASP Category	Primary CWEs	Detection Approach
A01	Broken Access Control	CWE-200, CWE-201, CWE-352, CWE-639, CWE-862, CWE-863	SAST (authorization checks), code review (RBAC patterns)
A02	Cryptographic Failures	CWE-259, CWE-261, CWE-327, CWE-328, CWE-330, CWE-522	SAST (weak algorithms), secret scanning
A03	Injection	CWE-20, CWE-77, CWE-78, CWE-79, CWE-89, CWE-94	SAST (taint analysis), DAST (fuzzing)
A04	Insecure Design	CWE-209, CWE-256, CWE-501, CWE-522	Architecture review, threat modeling
A05	Security Misconfiguration	CWE-2, CWE-11, CWE-16, CWE-611	IaC scanning, configuration audit
A06	Vulnerable and Outdated Components	CWE-1104	SCA (Dependabot, Dependency Review), SBOM
A07	Identification and Authentication Failures	CWE-255, CWE-287, CWE-384, CWE-613	SAST (session management), DAST (auth flows)
A08	Software and Data Integrity Failures	CWE-345, CWE-426, CWE-494, CWE-502, CWE-829	CI/CD integrity checks, artifact attestations
A09	Security Logging and Monitoring Failures	CWE-117, CWE-223, CWE-532, CWE-778	Code review (logging patterns), monitoring audit
A10	Server-Side Request Forgery (SSRF)	CWE-918	SAST (URL construction), DAST (SSRF payloads)

OWASP LLM Top 10 (AI/LLM-Specific)

#	Category	CWEs	Relevance
LLM01	Prompt Injection	CWE-77, CWE-94	Agent configuration files, user input to LLM
LLM02	Insecure Output Handling	CWE-79, CWE-94	LLM-generated code execution
LLM03	Training Data Poisoning	CWE-506	Model supply chain
LLM04	Model Denial of Service	CWE-400	Resource exhaustion
LLM05	Supply Chain Vulnerabilities	CWE-829, CWE-494	Agent packages, prompt files
LLM06	Sensitive Information Disclosure	CWE-200	PII in training data or outputs
LLM07	Insecure Plugin Design	CWE-862	MCP tools, agent tool permissions
LLM08	Excessive Agency	CWE-269	Agent tool access scope
LLM09	Overreliance	N/A	Trust calibration
LLM10	Model Theft	CWE-200	Model exfiltration

ASP.NET Core Security Patterns

Authentication and Authorization

Pattern	Implementation	Risk if Missing
JWT validation	AddAuthentication().AddJwtBearer() with issuer/audience validation	A07 — Auth bypass
Role-based authorization	[Authorize(Roles = "Admin")] or policy-based [Authorize(Policy = "...")]	A01 — Broken access control
CORS configuration	AddCors() with explicit origins, never AllowAnyOrigin() with credentials	A01 — Cross-origin abuse
Anti-forgery tokens	[ValidateAntiForgeryToken] on state-changing actions	A01 — CSRF (CWE-352)
Rate limiting	AddRateLimiter() with fixed/sliding/token bucket policies	A04 — Abuse prevention

Input Validation

Pattern	Implementation	Risk if Missing
Model validation	Data annotations + [ApiController] automatic 400 responses	A03 — Injection
Parameterized queries	EF Core LINQ or Dapper with parameters, never string concatenation	A03 — SQL injection (CWE-89)
Output encoding	Razor auto-encodes by default; use HtmlEncoder for raw output	A03 — XSS (CWE-79)
File upload validation	Validate content type, size, and extension; store outside web root	A04 — Arbitrary file upload

Secrets Management

Pattern	Implementation	Risk if Missing
User Secrets (dev)	dotnet user-secrets for local development	A02 — Secrets in source
Azure Key Vault (prod)	AddAzureKeyVault() configuration provider	A02 — Hardcoded secrets
Managed Identity	DefaultAzureCredential with no connection strings in config	A02 — Credential exposure

Security Headers

Header	Value	Purpose
X-Content-Type-Options	nosniff	Prevent MIME sniffing
X-Frame-Options	DENY or SAMEORIGIN	Prevent clickjacking
Content-Security-Policy	Strict CSP with nonce/hash	Prevent XSS, injection
Strict-Transport-Security	max-age=31536000; includeSubDomains	Force HTTPS
Referrer-Policy	strict-origin-when-cross-origin	Control referrer leakage

IaC Security Checklist

Terraform

Check	Rule	CWE
No hardcoded secrets in .tf files	Scan for password, secret, key in plain text	CWE-259
Encryption at rest enabled	azurerm_storage_account.enable_https_traffic_only = true	CWE-311
Network segmentation	No 0.0.0.0/0 in NSG rules or firewall exceptions	CWE-284
Logging enabled	Diagnostic settings on all resources	CWE-778
TLS 1.2+ enforced	min_tls_version = "TLS1_2" on all services	CWE-327
Managed Identity over keys	Use identity {} block, avoid access_key references	CWE-522
State file encryption	Remote backend with encryption, never local state in CI	CWE-312

Bicep

Check	Rule	CWE
Secure parameters	@secure() decorator on secrets, never @description with default values	CWE-259
Resource locks	Microsoft.Authorization/locks on production resources	CWE-284
Diagnostic settings	Deploy Microsoft.Insights/diagnosticSettings with all resources	CWE-778
Private endpoints	Use Microsoft.Network/privateEndpoints for data services	CWE-284
Key Vault references	reference(keyVaultId).getSecret('name') instead of inline secrets	CWE-522

Kubernetes / Helm

Check	Rule	CWE
No privileged containers	securityContext.privileged: false	CWE-250
Read-only root filesystem	securityContext.readOnlyRootFilesystem: true	CWE-732
Resource limits	CPU and memory limits set on all containers	CWE-400
Network policies	Default-deny NetworkPolicy with explicit allow rules	CWE-284
No latest tag	Pin image tags to specific digests or versions	CWE-829
Secrets from external store	Use CSI Secret Store driver or sealed secrets, never kind: Secret	CWE-312
Pod Security Standards	Enforce restricted profile via PodSecurity admission	CWE-250

CI/CD Pipeline Hardening Checklist

Domain	Check	Implementation
Secrets	No secrets in workflow files	Use ${{ secrets.NAME }}, never inline values
Secrets	Rotate secrets regularly	Automated rotation via Key Vault or GitHub secret scanning
Permissions	Least-privilege permissions	Set permissions: read-all at workflow level, escalate per job
Dependencies	Pin actions to SHA	uses: actions/checkout@abcdef1 not @v4
Dependencies	Enable Dependabot for actions	dependabot.yml with package-ecosystem: github-actions
Isolation	Use ephemeral runners	GitHub-hosted or self-hosted with clean images per run
Artifacts	Sign and attest artifacts	actions/attest-build-provenance for SLSA provenance
Artifacts	Verify artifact integrity	Check SHA256 digests before deployment
Branch protection	Require status checks	Branch rules enforce CI pass before merge
Branch protection	Require code review	Minimum 1 approval, dismiss stale reviews
SARIF upload	Upload all scanner results	github/codeql-action/upload-sarif@v4 per domain
Audit	Log all workflow runs	GitHub audit log + Defender for Cloud integration

Supply Chain Security Controls

Control	Tool	Purpose
Secret scanning	GitHub Secret Protection + custom patterns	Prevent credential leaks in commits
Dependency scanning	Dependabot + Dependency Review	Detect vulnerable dependencies before merge
SBOM generation	Anchore Syft, Microsoft SBOM Tool	Inventory all components
Container scanning	Trivy, Grype, MSDO	Detect vulnerabilities in container images
Artifact attestation	actions/attest-build-provenance	SLSA Build Level 2+ provenance
Code signing	Sigstore / Cosign	Verify artifact authenticity
Agent config scanning	APM apm audit	Detect hidden Unicode in agent files
Lock files	package-lock.json, requirements.txt with hashes	Prevent dependency confusion

Output Format for Security Findings

Read the security-reviewer-formats skill for full format specifications before generating any report. Follow its normative reference links to load:

• Report Formats (references/report-formats.md) — VULN_REPORT_V1 template (audit and diff modes), diff mode qualifiers, and PLAN_REPORT_V1 template (plan mode).
• Finding Formats (references/finding-formats.md) — Verified Findings Collection Format describing the input structure.
• Completion Formats (references/completion-formats.md) — Scan Completion Format used by the orchestrator after report delivery.