What is sail-skill?

sail-skill packages a security framework called SAIL V2, created by a company named Pillar Security, into a form that AI coding assistants and agents can use directly. SAIL stands for Secure AI Lifecycle, and the project bundles a catalog of 91 known risks that AI systems and autonomous agents can run into, spread across 7 stages of an AI system's life, each with suggested fixes and links to relevant industry standards. Once installed, the skill activates automatically whenever you are discussing AI or agent security with your assistant. You can ask it to review the security of an agent you built, describing its parts and how they connect, and it will check each of the 91 known risks against your design. It can also build a step by step security roadmap for a team that has agents running but no formal policy yet, produce compliance checklists mapped to standards such as the EU AI Act, ISO/IEC 42001, or OWASP's guidance for AI systems, and draft the security related sections of a vendor request for proposal document. The project supports many different AI tools, including Claude Code, OpenAI's Codex, Google's Antigravity, ChatGPT, and other tools that support the shared SKILL.md format for extending an assistant's abilities. Claude Code users get extra shortcut commands for launching an assessment, generating a roadmap, building a compliance checklist, or drafting a vendor questionnaire. Installation methods vary by tool: Claude Code supports a one line plugin install, while other tools use their own install command or a manual copy of the skill folder into a specific directory. The project positions itself as grounding its answers in the actual catalog of risks rather than producing generic or invented security advice.