Cratory — Invoicing & Multi-Tenant Finance Platform
• Engineered an Automated Double-Entry Engine: Programmed background ledger logic that automatically records dual debit/credit journal entries upon Sales Invoice generation, Purchase Bills, or...
An open-source security skill and automated scanner for AI coding assistants (Claude Code, Antigravity, Cursor, Windsurf) to catch vulnerabilities in AI-generated code.
Engineered with clean architectural discipline, efficient relational queries, and resilient user interface bindings to guarantee high performance and uptime.
Structured controllers and service abstractions ensuring maintainable enterprise business logic.
Indexed relational structures and normalized tables preventing bottlenecks under concurrent load.
Input sanitization, parameterized queries, and strict role permissions across all endpoints.
AI coding assistants have accelerated feature delivery across modern software teams, but they systematically produce code with subtle, critical security blind spots. Under pressure to make code work, LLMs produce decorative authorization checks, unpinned JWT verifiers accepting alg:none, mass assignment vulnerabilities (User(**request.json)), and unvalidated SSRF fetch wrappers reaching cloud metadata (IMDSv1). These flaws quietly pass unit tests and human code review because the code looks structurally correct, leaving production systems exposed from day zero.
I architected and open-sourced Day Zero Security (v1.4.0), a domain-organized security methodology following the open Agent Skills standard (agentskills.io). To eliminate vulnerabilities at code-generation time, the skill introduces: • 10 Security Domains & 76 Detection Categories: Comprehensive coverage spanning OWASP Top 10 (2021), OWASP API Security (2023), Mobile Security (iOS/Android), Cloud & IaC misconfigurations, and LLM supply-chain risks (slopsquatting, tool poisoning). • Zero-Dependency Regex Baseline: A portable bash scanner (scripts/scan.sh) running 340+ ERE regexes in seconds using ripgrep or POSIX grep fallback without third-party dependencies. • AST & Taint Flow Tracking: 32 Semgrep detection rules providing taint-confirmed source-to-sink vulnerability verification. • Dual Mode Architecture: Mode A for auditing legacy applications with formal attack-chaining rubrics, and Mode B for guiding AI agents to scaffold secure-by-default architectures before writing logic. • Automated CI Self-Audit Pipeline: 76 automated regression checks guarding detection accuracy and pattern portability on every commit.
Day Zero Security was published on GitHub as an open-source tool (MIT license). The repository features 100% green passing CI workflows across 5 automated test jobs, a complete multi-agent adapter suite (supporting Claude Code, Antigravity, Cursor, Windsurf, Kiro, and Copilot), and an attack-chaining procedure that re-rates vulnerabilities based on full exploit path rather than isolated symptoms. It is now freely available for developers and DevSecOps teams to embed directly into their daily AI coding workflows.
• Engineered an Automated Double-Entry Engine: Programmed background ledger logic that automatically records dual debit/credit journal entries upon Sales Invoice generation, Purchase Bills, or...
• Scalable Relational Schema: Redesigned the employee and attendance relational schema in MySQL, introducing composite indexing and eager-loading to eliminate N+1 query bottlenecks. •...