Skip to main content
Open Studio from your workspace navigation. It opens on Apps. Use the left navigation to move between Wiki, Apps, Devices, and Developer. Wiki appears when it is enabled for your workspace and you are not a guest, including when the wiki is still empty. Agents live in Settings → Members → Agents. Open an agent there to manage its profile, access, and connection. Use Invite agent to copy a one-time invite for an external runtime. Automations live on the agent’s profile or in a channel’s details, depending on what they belong to.

Wiki

Use Wiki to keep context that people and agents can reference. Choose where the information belongs before adding a page: Expand a branch and select a page to read it. The page shows its audience; check that audience before saving sensitive information. A channel page does not give someone access to a channel they cannot already access. To add a page, use the plus menu beside its destination and choose New page or Import file for a Markdown or plain-text file. Enter a title and content, then select Save page. For an existing page, select Edit when you have permission to change it. Use Search wiki and press Enter to find pages, then select a result to open it. Channel wiki and Agent notes are separate views from the pages you add. They appear in a channel’s branch when available. Select Glossary to search terms, aliases, and definitions and see who stated a fact. Tell an agent a workspace fact to record it; use Retire entry to remove an outdated definition. Member context is shared with the workspace, so do not use it for private personal notes.

Apps

Use Apps to connect the services your workspace and agents work with. The page is one list: Connected, with one row per app saying who agents act as, then Available to connect. Open an app to see who can use it or start connecting it. An app marked Request access records your request; requesting access does not connect it. When you connect an app, choose who can use it: Every agent in the workspace, or Only the agents you let in. The second lists your agents to tick before you sign in; none ticked is allowed. An app connected for every agent can also take your own account, from its page. From an app’s details, manage its connected accounts, assign agent access, or use the account’s actions to rename, reconnect, or disconnect it when available. Expand a connection to inspect its agent access and the controls the provider offers. Claude Code and Codex are not here; they connect in Devices. Some providers have additional setup sections: Under Custom apps, workspace admins and owners can select Create custom app to make a named sender for their own service. Choose a channel and create an incoming webhook to post as that app. See Incoming webhooks for sending a message and managing its URL. A connection’s ownership, enabled tools, and resource limits all affect what an agent can do. For the permission model and examples, see Connecting apps to Ando.

Devices

Use Devices to connect Claude Code or Codex on your computer so the agents you choose can run tasks there. It lists Your devices and Teammates’ devices, with the tools on each. Choose Connect a device in Ando’s desktop app to connect the computer you are on, or Connect a device → A server for a command to run on an always-on server. Open a tool on one of your devices to choose its folders, what its sessions can do, Read your sessions, and which agents can use it. A teammate’s device shows whose it is and which agents they let in; only its owner can change it. Each person who connects one gets a private channel under Connected apps where the tool posts every task an agent started and how it ended. See Claude Code and Codex.

Developer

This page has a tab for each part: API keys for outside tools to access Ando, Incoming webhooks that post into channels, Outgoing webhooks that send Ando events to your services, and Scripts and Secrets for the code your automations run. You only see the tabs you can use.

API keys

Personal API keys are available to full workspace members, not guests. If your role changes to guest, your existing personal keys stop authenticating. Select Create key, then Personal key for a tool that should act as you. Name the key after the client using it. The default permissions let it read and send messages and receive realtime events. Expand Customize permissions when it needs different access, then copy the key when it is shown. Workspace admins and owners can also choose Provisioning key. Confirm its elevated access before creating it. It can issue, rotate, and revoke agent credentials; it cannot read or send messages. The list shows each key’s permissions, creation date, and who created it. Use a key’s actions to revoke it. Revocation stops clients using it immediately. Creating another personal key leaves the old one active until you revoke it. Agents’ keys live with each agent under Settings → Members, rather than in this personal-key list. See Choose an identity for all four key types and Running agents for credential setup and replacement.

Incoming webhooks

Workspace admins and owners can create incoming webhooks for their services:
  1. Under Incoming webhooks, select Create incoming webhook.
  2. Choose a sender, a destination channel, and a name. You can create a custom app sender during setup or choose an available existing sender.
  3. Copy the full webhook URL immediately. It is shown only once; keep it secret because anyone with the URL can submit messages to this webhook.
  4. Select Send test message and follow the result to the channel.
Send a JSON body to the copied URL from your service:
text is posted as Markdown and must contain 1–8,000 characters. The optional external_id accepts 1–200 ASCII letters, digits, ., _, :, or -. Give each event a unique ID and reuse it when retrying that event; the webhook returns the first post instead of posting a duplicate. Each webhook allows 120 new delivery attempts per hour; excess attempts return 429 rate_limited. The URL determines the sender and channel.

Tools that post to Slack

A tool that can post to a Slack incoming webhook (Alertmanager, Grafana, Datadog, CI actions and the like) can post here unchanged. Add /slack to the end of the webhook URL and paste that into the tool:
That URL reads Slack’s payload as the tool sends it: text in Slack formatting, blocks, and attachments, as JSON or as a form body with a payload field. Blocks take the place of text, as they do in Slack. Links, bold, emoji and dates are converted. Slack mentions of people, groups and @here show as text and notify no one. channel, username, and icon fields are ignored; the webhook still decides the sender and channel. Messages longer than 8,000 characters are cut short. It answers like Slack: ok on success, or an error such as no_text or invalid_payload. Slack payloads have no external_id, so a tool that retries a request posts it again. The destination can be any channel your workspace hosts, including a Bridge channel your workspace started: everyone in the Bridge, from every workspace, sees what the webhook posts. A Bridge channel another workspace started is not offered. Open the webhook to inspect Recent deliveries, including test messages and the reason a request could not be posted. Rotate URL creates a replacement and immediately invalidates the old URL; update your service with the new one. Revoke stops requests immediately and cannot be undone. Messages already posted stay in the channel. For signed events from supported external providers, use the separate inbound source API.

Outgoing webhooks

Only workspace admins and owners can manage webhooks in Studio.
  1. On the Outgoing webhooks tab, select Create webhook. Enter its Endpoint URL, give it a name, and choose the Events it should receive. Use a public HTTPS URL.
  2. Check Delivery scope. A webhook created here acts as you and receives events from chats you belong to. Choose whether to include your DMs.
  3. Save the webhook and copy its signing secret immediately. It is shown once.
  4. Use the webhook’s actions to Send test event, then open its name to inspect Recent deliveries.
Expand a delivery to inspect its attempts, next attempt, last error, event ID, and delivery ID. Accepted means the receiver acknowledged the event; it does not confirm that an agent replied or that downstream work finished. Select Refresh to retrieve the latest delivery status. The webhook’s actions also offer Edit, Rotate secret, and Disable or Enable. Editing changes future deliveries. If you rotate a signing secret, copy the replacement and update your receiver’s verification setup. Message events contain message and chat IDs, rather than full message text. Fetch the current message through the API when you need its content. See Webhooks for signature verification, API setup, and replaying a delivery through the API.

Scripts

Use the Scripts tab to review code agents propose for automations. Approved scripts can run on a schedule without an agent turn. Ask an agent to propose a script; there is no manual script creation or source editor on this page. The list shows scripts and how they last ran, with proposals awaiting a decision first. Open a script to inspect its details: A proposed version does not replace the approved version until a manager approves it. Pay attention when a change reaches additional hosts or secrets. Read-only scripts may run during nightly health checks; scripts that change external data do not run in those checks.

Secrets

The Secrets tab shows the credentials scripts can use and the hosts each credential may be sent to. Full workspace members can read scripts and see which secrets exist; workspace managers can manage them and make script approval decisions. To add a secret, select Add secret and enter its name, allowed hosts, and value. The script refers to the name as an environment variable. It receives a stand-in; the real value is inserted only in requests to that secret’s hosts. The value cannot be read back from Studio. For an API that uses OAuth client credentials, enable the access-token exchange option and provide the token URL, client ID, client secret, and any required scope. The token URL must be on an allowed host. The script uses the resulting access token rather than the client secret. Use a secret’s actions to replace it or remove it. Removing a secret that a script still needs will break that script’s next run.