> ## Documentation Index > Fetch the complete documentation index at: https://docs.kanaeru.ai/llms.txt > Use this file to discover all available pages before exploring further. # Settings and profile > Where each setting lives: workspace settings, your profile, connectors, and automations. Kanaeru keeps settings in a few places: | Where | How to get there | What's there | | - | - | - | | **Workspace Settings** | A workspace's **…** menu → **Settings** | Stats, connected services, context sources, the workspace's automations, members, agent memory, agent permissions, autonomy, archived workspaces | | **Profile** | User menu (your name at the bottom of the sidebar) → **Profile** | Your account, usage, reading helper, personal GitHub, connectors, sign out | | **Settings** | User menu → **Settings** | Theme and language | | **Automations** | User menu → **Automations** | The current workspace's settings page, where its automations are. See [Automations](/guides/automations) | | **Connectors** | `/settings/connectors` | Connected services and runtime MCP servers. See [Connectors](/guides/connectors) | | **Machine page** | **Connect** in the chat box's runtime selector, or `/machine-connect` | Machines, runtimes, provider credentials, harness bundles. See [Connect a machine](/machines/connect-a-machine) | ## Workspace Settings **Workspace Settings** opens with an overview of the workspace: its outcomes, current phase, connected services, linked items, and member count. Below that: * **Connected Services** and the workspace's **context sources**. See [Connectors and context](/guides/connectors). * The workspace's **Automations**. * **Members**: how many people are in the workspace. * **Agent Memory**, **Agent Permissions**, and **Autonomy by role**, described below. * **Archived workspaces**, where owners and admins restore a workspace that was archived. You can also open **Archived workspaces** from the sidebar and choose **Restore workspace** from a workspace's **…** menu. Owners and admins of your Kanaeru organization control permissions and autonomy. They also manage company notes and review learnings. ## Agent permissions **Agent Permissions** sets what agents may do and what they may spend. * **Grants:** whether agents may **Read**, **Write**, and **Send** on your behalf. * **Monthly spend cap:** a limit in USD for the whole organization, and a separate **Monthly spend cap per user**. `0` means unlimited. By default every grant is on and there's no cap. Both caps are checked before every chat turn, on every machine, whether it's a cloud machine or your own computer. When the organization or a person reaches a cap, new turns can't start until the cap is raised or the month resets. A turn that's already running when the cap is crossed finishes. ## Autonomy **Autonomy by role** sets how much review each kind of work gets. Each role is set to one of five levels: | Level | Meaning | | - | - | | **Human only** | Only a person does this work | | **Approve each** | Kana asks before each action | | **Review first** | Kana prepares the work and you review it before it takes effect | | **Act & notify** | Kana acts and tells you afterwards | | **Autonomous** | Kana acts without checking in | Roles are Vision, Planning, Refinement, Review, Task execution, Koa operator, Retro, and Artifact review. Each role has a floor and a ceiling, and the level within that range is earned from Kana's track record. ## Agent memory **Agent Memory** shows **Saved learnings** for the workspace, and **Company notes**: key/value entries for your organization, such as your API style or the team's naming rules. Kana's Planning, Task Execution, Review, Vision & Roadmap, and Refinement sessions read company notes as company guidance, so only owners and admins can add, edit, or delete them; everyone else sees them read-only. See [The learning loop](/concepts/learning-loop). Kana also keeps **learnings**, lessons drawn from retrospectives and earlier work. If you're an owner or admin, you manage them in **Agent Memory**: * **Pending review** lists company-wide learnings agents have proposed. Open one to read its evidence, then **Approve** or **Dismiss** it with a reason. Nothing reaches future sessions until it's approved. Dismissing also retires the same lesson everywhere it was saved in the company. If the learning changed while you had it open, you're shown the latest version to review again. * **Saved learnings** shows the workspace's learnings. On a workspace or outcome learning you can **Resolve** it (the problem is fixed; it comes back if the problem recurs), **Dismiss** it (the advice is wrong), or **Reactivate** a resolved one. Each needs a reason and changes only that learning. The same actions are available from the terminal: ```bash theme={null} kana memory patterns kana memory patterns dismiss --reason "Doesn't apply to this repository" --yes kana memory learnings list kana memory learnings approve --revision --reason "Matches our API style guide" --yes ``` ## Profile Your **Profile** shows your account (email, role, company, when you joined, and how you sign in) and how many sessions you've run. It's also where you: * turn on **Reading helper**, which underlines Kanaeru terms and explains them inline. It's a personal preference, so it doesn't change anything for your teammates. * choose **Connect personal GitHub**, so work in your own Kana sessions uses your verified GitHub identity, only within the repositories your company's Kana GitHub App can reach. See [GitHub](/guides/github). * connect services. Owners and admins also install the Kana GitHub App here. * **Sign out**. Roles in your organization are **owner**, **admin**, and **member**. ## Language and appearance Kanaeru is available in English and Japanese. Change the language and theme from **Settings** in the user menu. Turn on notifications from the chat page. ## Related topics - [GitHub](/guides/github.md) - [Session plugins](/machines/session-plugins.md) - [Connectors and context](/guides/connectors.md) This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.