# Silvector Docs > Built for the hardest cases. Made for everyone. Silvector decides what to do with each request from software and people, signs a receipt for every decision, and grades the evidence on a six-rung continuum (0 Intuition to 5 Evidentiary). Receipts are Ed25519 over canonical JSON; keys at https://api.silvector.com/.well-known/jwks.json. ## Overview - [Silvector Docs](/): Silvector documentation, organized by job and by role: quickstarts to a verified signed receipt, the detection continuum, concepts, playground and API reference. ## Get started - [Quickstarts](/quickstarts/): Reach a verified, signed Silvector receipt in under ten minutes with Cloudflare Workers, Node, Python or Go. - [Decision playground](/playground/): Send a request shape to Silvector, get a decision and a signed receipt, and verify it in your browser. ## Quickstart - [Cloudflare Workers quickstart](/quickstarts/workers/): Verify a signed Silvector receipt with Cloudflare Workers in under ten minutes. - [Node quickstart](/quickstarts/node/): Verify a signed Silvector receipt with Node in under ten minutes. - [Python quickstart](/quickstarts/python/): Verify a signed Silvector receipt with Python in under ten minutes. - [Go quickstart](/quickstarts/go/): Verify a signed Silvector receipt with Go in under ten minutes. ## By job - [Jobs](/jobs/): Task paths through Silvector, from stopping card testing to building a case that holds up. - [Stop card testing by agents](/jobs/card-testing/): Slow down automated card testing at checkout without declining real customers. - [Let AI shopping agents buy safely](/jobs/shopping-agents/): Admit agents that can show who they are and who they act for, and step up before money moves. - [Show an auditor what your policy did](/jobs/auditor/): Hand over records an auditor can check without trusting you or Silvector. - [Check whether this is AI without accusing anyone](/jobs/check-without-accusing/): Get an honest signal, understand what it cannot tell you, and protect the person on the other side. - [Build a distillation or impersonation case that holds up](/jobs/distillation-impersonation-case/): Collect, preserve and present a distillation or impersonation case so another expert can check every step. ## By persona - [Personas](/personas/): Where to start in the Silvector docs, by role. - [CISO and head of security](/personas/security/): Agents hit login, checkout and APIs. You cannot tell good from bad, and a block-only bot vendor cannot explain itself. - [Fraud and trust lead](/personas/fraud/): Account takeover, card testing, scalping and promo abuse now run through agentic tooling. - [Head of growth and e-commerce](/personas/growth/): AI shopping agents are a new channel, and it is being blocked without anyone deciding to block it. - [Platform and SRE lead](/personas/platform/): Scraping and agent load cost money, and the rules that manage it are noisy and unexplained. - [Product, legal and compliance](/personas/compliance/): Content licensing, training scrapers, the EU AI Act and audits all ask what your policy actually did. - [AI agent builders](/personas/agent-builders/): Good bots want a reliable, signed way in instead of being guessed at. - [Frontier lab trust and safety, and model owners](/personas/frontier-labs/): Unauthorized distillation, model extraction, account farms and espionage. - [Investigators, law enforcement and litigators](/personas/investigators/): Deepfake impersonation, fraud rings and evidence that must survive a court. - [Newsrooms and platform integrity](/personas/newsrooms/): Synthetic media and coordinated campaigns, with the risk of a wrongful takedown on the other side. - [Educators, HR and everyday users](/personas/everyone/): You want to know whether something is AI without wrongly accusing a real person. ## Detection continuum - [The detection continuum](/continuum/): Six rungs from a hunch to evidence for a court: what each can claim, what it cannot, its error rates, and how to escalate. - [Rung 0: Intuition](/continuum/rung-0/): Is this AI? What rung 0 evidence can and cannot claim, its error rates, and how to escalate. - [Rung 1: Screening](/continuum/rung-1/): Which items need a closer look? What rung 1 evidence can and cannot claim, its error rates, and how to escalate. - [Rung 2: Operational](/continuum/rung-2/): What do I do with this request now? What rung 2 evidence can and cannot claim, its error rates, and how to escalate. - [Rung 3: Investigative](/continuum/rung-3/): Who is behind this series of events? What rung 3 evidence can and cannot claim, its error rates, and how to escalate. - [Rung 4: Forensic](/continuum/rung-4/): Was our model distilled, or our brand impersonated? What rung 4 evidence can and cannot claim, its error rates, and how to escalate. - [Rung 5: Evidentiary](/continuum/rung-5/): Will this hold up? What rung 5 evidence can and cannot claim, its error rates, and how to escalate. ## Concepts - [Concepts](/concepts/): Precise definitions of principal, agent, delegation, evidence, decision, receipt, policy and tenant. - [Principal](/concepts/principal/): The accountable human or organization on whose authority something acts. - [Agent](/concepts/agent/): Software that sends requests, on its own or for a principal, and may carry a signature claim. - [Delegation](/concepts/delegation/): The on-behalf-of chain that binds an agent to a principal for a scoped task. - [Evidence](/concepts/evidence/): Records that let someone else check a claim, graded by how much weight they can bear. - [Decision](/concepts/decision/): The outcome a policy gives one request, with the rule and the facts behind it. - [Receipt](/concepts/receipt/): A signed record of one decision that anyone can check offline. - [Policy](/concepts/policy/): The rules and flag state that turn a request into a decision, identified by hash. - [Tenant](/concepts/tenant/): The isolation boundary: one organization's policies, receipts and data. ## Reference - [API reference](/api/): Reference for the Silvector public API: /v1/check, the receipt key set, the transparency log head and health. - [Demos](/demos/): Live, recorded and simulated Silvector demos, each labeled by how real it is. - [Changelog](/changelog/): Changes to the Silvector public API, receipts, keys and transparency log. - [Status](/status/): Current health of the Silvector public API and edge, checked from your browser. ## Trust - [Trust and security](/trust/): How Silvector signs, stores and limits what it records, what it does not claim, and how to report a vulnerability.