> ## Documentation Index > Fetch the complete documentation index at: https://docs.cloudthinker.io/llms.txt > Use this file to discover all available pages before exploring further. # How CloudThinker Fits Together > See how organizations, workspaces, connections, agents, guardrails, and modules fit together before you set anything up. CloudThinker has a small number of building blocks: an organization holds workspaces, and each workspace holds the connections, agents, guardrails, and context its modules run on. This page shows how they fit, and links to each one. ## The model at a glance CloudThinker model: an organization holds workspaces; each workspace has connections, agents, guardrails, and context that power the four modules, with results in Home, Investigation, and Artifacts CloudThinker model: an organization holds workspaces; each workspace has connections, agents, guardrails, and context that power the four modules, with results in Home, Investigation, and Artifacts ## Organization and workspaces Your [organization](/guide/organization) is the account: it owns billing, the plan, and the member list. A [workspace](/guide/workspaces) is an isolated environment inside it — production and staging, or one per team. | Lives in the organization | Lives in each workspace | | - | - | | Billing, plan, and credits | Connections and credentials | | Members and org roles: Owner, Admin, Developer, Viewer | Custom agents | | Single sign-on and audit logs | Memory, knowledge bases, and skills | | | Automations, commands, and approval settings | Org Owners and Admins are Admins in every workspace. Everyone else gets a workspace role — Admin, Developer, or Viewer — per workspace. See [Workspace users](/guide/workspace-users). ## Connections A [connection](/guide/connections/overview) is a credential for one service — an AWS account, a Kubernetes cluster, a GitHub organization, a Datadog account. Agents can only see and act on what the workspace's connections allow, so the connection's permissions are the outer limit of what any agent can do. ## Agents **CloudThinker**, the built-in assistant, takes every request. It spawns temporary subagents for parallel work and hands off to custom agents your team defines. See [Agents](/guide/agents/overview). ## Guardrails Two settings decide whether an agent's write action runs on its own or waits for a person. | Setting | What it does | Where | | - | - | - | | **Auto** (default for new workspaces) | Write actions are checked by an AI safety classifier before they run | **Approval** control in the chat box | | **Manual** | You approve each write action before it runs | **Approval** control in the chat box | | **Tool Permissions** | In Manual, set each connection tool to **Always allow**, **Needs approval**, or **Never allow** | **Settings → Approval → Tool Permissions** | The mode applies to the whole workspace, including automations and subagents. See [Auto Mode](/guide/auto-mode) and [Approval](/guide/approval). ## Context Context is what agents know before you say anything. Each kind has a different source. | Context | What it holds | Who adds it | | - | - | - | | [Memory](/guide/memory) | Notes about your environment, preferences, and recent events | Agents, as they work; you can correct it | | [Knowledge](/guide/knowledge) | Your runbooks, policies, and documentation | You, by uploading files or crawling sites | | [Skills](/guide/skills/overview) | Step-by-step instructions for a specific kind of task | You, or agents when skill learning is on | ## Modules The four [modules](/guide/modules) — Resolve, Review, Optimize, and Cyber — each apply the same agents, guardrails, and context to one kind of signal: incidents, pull requests, cloud bills, and app security. ## Where results appear | Place | What you find there | | - | - | | [Home](/guide/activity) | **Needs you** — approvals and decisions waiting on you — plus module status and today's automations | | **Investigation** (under Resolve) | Every incident, its root cause, evidence, and proposed fix | | [Artifacts](/guide/artifacts/overview) | Dashboards, reports, and files agents built for you | | The chat side panel | Runs, console output, subagents, and findings for the current conversation | ## Related Learn how CloudThinker, subagents, and custom agents share the work Compare what each module does and what it needs Create and organize isolated environments Put the model to work in about 10 minutes This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.