What problem does it solve? Code changes that touch auth, secrets, untrusted input, SQL, shell execution, or dependencies carry security risk that is easy to miss in fast-moving AI-assisted workflows. This Skill routes those changes through the Rafter CLI so scans, command-risk checks, and finding triage actually happen before work is called done. ## Core Features & Use Cases - Remote and local scanning: Run rafter run for remote SAST + SCA + secrets analysis, rafter run --mode plus for an agentic deep-dive, or rafter secrets for offline hardcoded-credential detection with no API key. - Command risk evaluation: Classify shell commands into low/medium/high/critical risk tiers with rafter agent exec --dry-run before executing destructive operations. - Finding triage and suppression: Interpret severity, confidence, and rule IDs, then fix, baseline, or suppress findings with documented reasons in .rafter.yml. - Use Case: Before merging a PR that adds a new API endpoint and bumps dependencies, run rafter run to catch injection flaws and known-vulnerable packages, then triage any findings by severity and confidence. ## Quick Start Ask the AI to run a Rafter security scan on the current repository and summarize any findings by severity.