Getting Started
Getting Started is the orientation path through SaferSkills. In four short pages you learn what the service is, why independently scanning AI capabilities matters, how to install a verified capability in about five minutes, and the core concepts the rest of the docs build on. Read them in order if SaferSkills is new to you, or jump to whichever question you came with.
What should you read first?
Section titled “What should you read first?”Read what SaferSkills is first — it explains the service in one screen: the deterministic ~30-second scan, the public report, the five capability kinds, and the difference between a static component scan and a behavioral Agent Scan.
What does each page in this section cover?
Section titled “What does each page in this section cover?”- What is SaferSkills? — the one-screen overview: “every AI capability, independently scanned,” the no-LLM-in-the-verdict-path guarantee, and what gets scanned.
- Why scanning matters — the threat model: agents run with your credentials and your shell, prompt injection is the top LLM risk, and vendor self-attestation is not independent verification.
- Quickstart — install a verified capability end to end and read its score before it touches your machine.
- Core concepts — a compact map of the five capability kinds, the 0–100 scoring model, and the trust model that ties them together.
Where do you go next?
Section titled “Where do you go next?”After this section, the Concepts pages go deeper on each capability kind, Find & Verify shows you how to use the running service, and Security & Methodology documents the scoring model in full. Unfamiliar terms are defined in the glossary.