with one click
railguard-skill
railguard-skill contains 6 collected skills from Netflix-Skunkworks, with repository-level occupation coverage and site-owned skill detail pages.
Skills in this repository
Store security scan findings to SQLite and compare against answer sheets or other scan runs. Use when asked to "store findings", "save scan results", "compare against answer sheet", "benchmark scan", "compare scan runs", "how did the scan perform", "evaluate scan accuracy", "check recall", "show dashboard", "benchmark dashboard", "benchmark trends", or "leaderboard".
Multi-phase security scanner using 12 parallel specialized subagents with SQLite progress tracking. Findings persist to DB for compaction resilience. Best for large repos. Use when asked to "db-tracked scan", "resilient scan", or "large repo security scan".
Security-agents-only scanner using all 12 parallel specialized subagents without discovery, data flow tracing, false positive analysis, or triage. Use when asked to "agents-only scan", "raw agents scan", "no-FP scan", or "security agents only scan".
Multi-phase security scanner using a single SuperAgent that loads all 12 rule files and covers all vulnerability domains in one pass. Lower cost alternative to the orchestrated scan. Use when asked to "superagent scan", "monoprompt scan", or "cheap security scan".
Multi-phase security scanner combining file-handoff parallel FP batching with SQLite progress tracking. Phase 3 agents write findings to /tmp files and return lean stubs. merge-findings.py batches findings for parallel Phase 4 FP agents. scan-progress.py tracks phase lifecycle, agent dispatch/skip/result, and enables crash recovery. Best of both orchestrated-scan and db-updates architectures. Use when asked to "unified scan", "tracked scan", or "resilient security scan". Supports Python, JavaScript, TypeScript, Java, Go, Ruby, and PHP.
Single-agent security scanner that groups vulnerabilities into 4 category sweeps (Injection, Access Control, Logic, and Network/CORS). Each sweep is gated by architecture detection. No subagent dispatch -- all analysis runs in one context window with sequential passes. Use when asked to "vulncategory scan", "category scan", "monolithic security scan", or "single-agent security scan". Supports Python, JavaScript, TypeScript, Java, Go, Ruby, and PHP.