npx agentspec-cli install e1674bb0-f69f-4099-8bd5-4becb3c7b87e --framework ClaudeCode→ .claude/skills/daniel/SKILL.md
# Daniel Comprehensive security analysis for application code and infrastructure with CMMC Level 2 compliance mapping. ## Workflow Routing | Workflow | When to Use | Output | |----------|-------------|--------| | ScanCode | Analyzing code for security vulnerabilities | Security analysis with vulnerability findings, CMMC mapping, remediation guidance | | PerformSTRIDE | Comprehensive threat modeling of feature | STRIDE threat analysis (all 6 categories) with priority rankings | | GenerateAudit | Creating CMMC compliance audit trail | Audit trail document with all vulnerabilities mapped to CMMC practices | | RunStandup | Multi-agent security review | Coordinated analysis from Daniel, Mary, Bob, and Murat | ## Examples ### Example 1: Scan code for vulnerabilities ``` User: "Scan this authentication code for security issues" Skill loads: Daniel → ScanCode workflow Output: Security analysis identifying SQL injection, weak passwords, missing MFA - Vulnerability: SQL Injection - String Concatenation - Severity: Critical - CMMC: SI.L2-3.14.6 (System and Information Integrity) - Mitigation: Use parameterized queries - Secure code example provided ``` ### Example 2: STRIDE threat modeling ``` User: "Perform STRIDE analysis on payment processing" Skill loads: Daniel → PerformSTRIDE workflow Output: Comprehensive threat model across 6 categories - Spoofing: 2 threats found - Tampering: 1 threat found - Repudiation: 1 threat found - Information Disclosure: 3 threats found - Denial of Service: 1 threat found - Elevation of Privilege: 0 threats - Priority ranking and timeline recommendations ``` ### Example 3: Generate CMMC audit trail ``` User: "Generate CMMC audit trail for security review" Skill loads: Daniel → GenerateAudit workflow Output: CMMC-compliant audit document - All vulnerabilities found - Mapped to CMMC practices (17 domains) - Assessor-ready format - Evidence of security analysis performed ``` ### Example 4: Multi-agent security standup ``` User: "Review authentication system with full team" Skill loads: Daniel → RunStandup workflow Output: Multi-perspective analysis - Daniel: Security threats (STRIDE + CMMC violations) - Mary: Business impact and user value - Bob: Timeline and capacity estimates - Murat: Test coverage and quality assurance - Synthesized decision with all perspectives ``` ## Capabilities ### Vulnerability Detection (50+ Patterns) **SQL Injection (10 patterns)** - String concatenation, template literals, ORDER BY, UNION, LIMIT - Stored procedures, second-order, blind, time-based, NoSQL **Cross-Site Scripting (10 patterns)** - Reflected, stored, DOM-based, event handlers - CSS, SVG, Markdown, JSON, meta refresh, JSONP **Authentication & Authorization (20 patterns)** - Missing authentication, hardcoded credentials, weak passwords - Rate limiting, MFA enforcement, IDOR, privilege escalation - Mass assignment, path traversal, file upload, JWT issues **CMMC Infrastructure (10 patterns)** - HTTP usage, missing security headers, CORS misconfig - Missing audit logs, backup, incident response, change control - Baseline configuration, vulnerability scanning, dependencies ### CMMC Level 2 Coverage (17 Domains) - **AC**: Access Control - **AT**: Awareness and Training - **AU**: Audit and Accountability - **CA**: Security Assessment - **CM**: Configuration Management - **CP**: Contingency Planning - **IA**: Identification and Authentication - **IR**: Incident Response - **MA**: Maintenance - **MP**: Media Protection - **PE**: Physical Protection - **PS**: Personnel Security - **RA**: Risk Assessment - **RE**: Recovery - **SA**: System and Services Acquisition - **SC**: System and Communications Protection - **SI**: System and Information Integrity ### STRIDE Threat Modeling - **S**poofing: Identity authentication attacks - **T**ampering: Data integrity attacks - **R**epudiation: Non-repudiation attacks - **I**nformation Disclosure: Confidentiality attacks - **D**enial of Service: Availability attacks - **E**levation of Privilege: Authorization attacks ## CLI Usage Daniel can also be invoked via command-line for CI/CD integration: ```bash # Scan single file daniel-scan src/auth/login.ts # Scan entire directory daniel-scan src/ # STRIDE threat modeling daniel-scan --stride src/payment.ts # JSON output for automation daniel-scan --json src/api.ts # Stdin support for pipes cat suspicious.js | daniel-scan --stdin ``` Exit codes: - `0`: No vulnerabilities found - `1`: Vulnerabilities detected ## Integration - Works with AgilePm skill (adds security requirements to user stories) - Works with TestArchitect skill (security test scenarios from findings) - Works with Security skill (STRIDE threat models, CMMC baselines) - Generates audit trails for CMMC assessors - Multi-agent coordination with Mary, Bob, Murat ## Test Coverage **Production Ready**: 100% test coverage (78/78 tests passing) - Acceptance: 13/13 (100%) - Critical Security: 31/31 (100%) - Authorization: 11/11 (100%) - CMMC Compliance: 23/23 (100%) ## Methodology Daniel follows security industry standards: - **STRIDE**: Microsoft's threat modeling framework - **OWASP Top 10**: Web application security risks - **CMMC Model v2.0**: DoD contractor compliance (110 practices) - **NIST 800-171**: Protecting Controlled Unclassified Information - **ATDD**: Acceptance Test-Driven Development for quality assurance ## Implementation Daniel is implemented in TypeScript with: - `src/daniel/security-review.ts` - Core vulnerability detection - `src/daniel/vulnerability-patterns.ts` - 50+ detection patterns - `src/daniel/cmmc-lookup.ts` - CMMC practice mapping - `src/daniel/stride.ts` - STRIDE threat modeling - `src/standup/orchestrator.ts` - Multi-agent coordination - `bin/daniel-scan.ts` - Command-line interface See `src/daniel/README.md` for detailed API documentation and usage examples.
Built on the Agent Skills standard
security-audit
Security guidance and vulnerability review for codebases, APIs, services, CLI tools, libraries, and daemons. Use for security questions, focused reviews, vulnerability research, security audits, or pen tests. Run the complete workflow only for explicit codebase audit or pen-test requests, full/comprehensive/end-to-end reviews, or requested report artifacts.
by AgentHub Crawler
AGENTS.md — SpiderShield
> Security scanner and runtime guard for MCP servers. > Static analysis, policy enforcement, DLP, prompt injection detection.
by AgentHub Crawler
CLAUDE.md
This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
by AgentHub Crawler
Sign in to leave a comment.
AGENTS.md
> Operating manual for LLM agents working with this DQMC codebase.
by AgentHub Crawler