> ## Documentation Index > Fetch the complete documentation index at: https://docs.rotascale.com/llms.txt > Use this file to discover all available pages before exploring further. # Introduction > RotaGrant states what an agent may do before it acts, refuses what exceeds that, and records what was decided either way. RotaGrant is a governance layer that sits beside the agents you already run. It does not replace your models, your framework, or your orchestration. It bounds what they are allowed to do. Before a consequential action, your agent asks whether it is authorised. Nine gates evaluate in a fixed order and return one of six outcomes. The answer arrives **before** the action, and the decision is recorded whether it was allowed or refused. ## The mental model A bounded authority a **named human** signed for: what this agent may do, up to what value, until when, under what conditions. Not a service account. Nine checks in a fixed order. Cheap structural ones first, so an action refused at gate 1 never reaches policy evaluation. Every decision, allowed or refused, sequenced and sealed. The refusals are the half that proves the control operated. ## What it is not These are the questions that end procurement calls, so they are answered here rather than three clicks in. **It does not evaluate models.** Evaluation tells you how a system scores on a benchmark. This states what an agent may *do*. Most teams need both, and they are different products. See [how this differs](https://rotascale.com/why/). **It does not make you compliant.** It assembles evidence, maps it to clauses, computes what it can, and names what it could not find. Whether that satisfies your obligations is regulatory interpretation and it rests with you and your counsel. That disclaimer sits inside the signed payload of every evidence pack, so removing it invalidates the signature. **It does not call an LLM.** Not for summaries, not for classification, not anywhere. A test fails the build if a model provider appears in the dependency graph. ## Where it runs Single-tenant, in your environment: your cloud, your data centre, or airgapped. No payload reaches Rotascale on any code path, and the deployment refuses to start if a setting selects a service outside your network. It also tells you which evidence property you gave up by disconnecting it. ## Where to go next Three lines in your agent's own code, and a refusal you can see. What each one asks, in the order they run, and why the order matters. Observe, shadow, canary, enforce. Nothing refuses until you say so. A running deployment, not a recording. Real agents, real refusals.