Account
AI agents and API keys
An AI agent such as Claude Code, Claude Desktop, Cursor or Codex can work on your guides for you: add spots from your map links and notes, fill in descriptions and properties, sort spots into chapters, add routes and places, upload your photos and draft the sales page. It connects to Sceniq's MCP server (the way agent apps reach outside tools) or to the REST API, and it signs in with an API key that you make in Settings.
You see its work in the studio as it goes. Putting a guide on sale is always your own click there. Most of a live guide's changes wait for a publish, but not all: see When buyers see the agent's work.
When an agent connects, the MCP server first hands it Sceniq's rules: never invent facts, never take photos from the web, ask before deleting anything, and put nothing live unless you ask.
What a key allows
A key works only on your own creator account. Its access is one of three:
| Access | What the agent can do |
|---|---|
| Read | See your guides, drafts included: their settings, spots, chapters, routes, places, photos, the sales page draft, what still blocks a publish, link health and your profile. Every key can read. |
| Write | Everything Read allows, plus create, change and delete guide content and photos, change a guide's settings, save the sales page draft, set prices, and change your display name and social links. A guide it creates starts as a draft. Some of these changes reach buyers at once (see below). |
| Publish | Everything Write allows, plus publish the changes of a guide that is already live, when you ask it to. Publish always comes with Write. |
In the key list these read Read only, Read and write and Read, write and publish.
What stays with you
No key can do these, whatever its access. They need you, signed in to the studio:
- putting a draft guide on sale (its first publish), or an archived guide back on sale;
- archiving a guide;
- publishing a sales page, taking it offline, rolling it back or deleting it;
- changing whether a guide is in the Sceniq Store, the store review request included;
- your sales and payouts: seeing them, setting up payouts, withdrawing;
- accepting a new creator agreement;
- your account itself, and making or revoking keys.
While a new creator agreement waits for you to accept it, price changes and publishes are refused, through a key as in the studio, until you accept it in the studio.
When buyers see the agent's work
The agent's changes follow the same rules as your own edits in the studio. A live guide's content waits for a publish; its settings and price, and your profile, are read live:
| What the agent changes | When buyers see it |
|---|---|
| Anything in a draft guide | Once you put the guide on sale, in the studio. |
| A live guide's content: its map area, spots, chapters and collections, routes and places, properties and photos | Once the guide's changes are published: by you in the studio, or by the agent with a Publish key when you ask it to. |
| A live guide's settings: its name, headline, thumbnail, country and store tags | At once, without a publish. |
| A live guide's price | At once: new purchases pay the new price. |
| A sales page draft | Once you publish that sales page in the studio. |
| Your display name and social links | At once, wherever they show: your profile page, your sales pages, and for your name, the guides buyers own. |
A live sales page can take up to 5 minutes to show a change.
API keys
The API keys box in Settings lists your active keys, newest first. With no keys at all it says so, and with none active it says there are no active keys. Keys start with sk_sceniq_.
Keys that stopped working stay on record, folded away under the list with their number beside the title: Revoked keys while all of them were revoked, Expired keys while all of them expired, and Revoked and expired keys once the list holds both. Click the title to open or close that list.
| Part of a key's row | What it shows |
|---|---|
| Name | The name you gave the key, then its first characters, so you can tell keys apart. The full key is never shown again. |
| Badge | Its access: Read only, Read and write or Read, write and publish. A key that stopped working reads Revoked or Expired instead. |
| The line under it | When the key was made and by whom, when an agent last used it, and the day it expires, if it does. |
| Revoke | Stops the key. Only active keys have it. |
New API key
New key, at the top right of the box, opens the New API key dialog.
| Field | What it does |
|---|---|
| Name | What you call the key, so you can find it in the list later. Name it after the agent or computer that will use it, for example "Claude Code on my laptop". |
| Access | Three boxes. Read is always ticked. Write starts ticked: untick it for a key that may only look. Publish starts unticked; ticking it ticks Write as well and holds it until you untick Publish. |
| Expires | Never (revoke manually), the default, keeps the key working until you revoke it. In 90 days and In one year stop it on that day. |
| Create key | Makes the key. It stays grey until the key has a name. |
Under the boxes, the dialog repeats what no key can do. Cancel closes it without making a key.
Your new key
After Create key, the dialog shows the new key, once. Copy puts it on your clipboard and reads Copied for a moment. Paste it into your agent's setup straight away (see Build with your AI agent). If copying fails, the dialog says so: select the key and copy it by hand.
Sceniq does not keep the key itself, so it can never show it to you again. Once the dialog closes, whether by Done, the X or the Escape key, the key is gone from the screen. If you lost it, revoke it and make a new one.
Revoking and expiry
Revoke asks first, in a browser prompt that names the key, because agents using it stop working immediately. Once you confirm, it is final: the agent's next request is refused, and the key cannot be turned back on. It moves into the folded list of keys that stopped working, marked Revoked, so you can still see what it was.
A key with an expiry date stops on that day and moves there too, marked Expired, and the folded list's title follows: Expired keys, or Revoked and expired keys next to a revoked key.
A key keeps working when you sign out of the studio. It stops when it is revoked or expires, when the person who made it is no longer on the creator account, and while the account is suspended. In each case the agent's requests are refused as if the key were wrong.
How many keys
An account holds up to 20 active keys. At the limit, Create key is refused with a note in the dialog: revoke a key you no longer use first. Revoked and expired keys do not count.
Build with your AI agent
This box in Settings, under API keys, has the setup for each agent app, in three steps:
- Make a key in API keys and put it in place of
sk_sceniq_YOUR_KEYin the snippet. - Paste the snippet into your agent's setup, as its tab says. The MCP server tells the agent the rules when it connects; the skill file adds Sceniq's full playbook for agents.
- Ask your agent to build a guide from your notes, map links and photos. Review every spot in the studio, then publish.
Each tab but curl carries its app's logo. A tab holds one app's snippet, a line on where it goes, and a Copy button at the snippet's top right.
| Tab | Where the snippet goes |
|---|---|
| Claude Code | Run it in a terminal, in the folder you work in. The first command adds Sceniq's MCP server to Claude Code. The key goes into your own Claude Code settings, not into the project; add --scope user to use Sceniq in every folder. The last two lines install the authoring skill, so Claude follows Sceniq's rules for guides. |
| Claude Desktop | In Claude Desktop, open Settings, Developer, Edit Config, add the snippet to claude_desktop_config.json and restart Claude. Claude Desktop's own connectors expect a sign-in rather than a key, so the snippet starts mcp-remote, a small open-source program that passes your key on. It needs Node.js on your computer. |
| Cursor | Add it to ~/.cursor/mcp.json, your own settings for every project. Keep the key out of a project's .cursor/mcp.json, which is often shared through git. |
| Codex CLI | Add it to ~/.codex/config.toml. Codex reads the key from the SCENIQ_API_KEY environment variable: set it as the snippet's last line shows. |
| curl (REST) | Plain web requests, for scripts of your own: every operation is also a JSON endpoint, and photos upload as raw bytes. |
Under the tabs, the box gives the MCP server's address, https://sceniq.earth/api/mcp, the REST API's base, https://sceniq.earth/api/v1, and a link to the developer docs.
From a guide's Spots page
On a guide's Spots page, the card Work on your spots with your AI agent has a Connect your agent button. It opens the same setup for that one guide, makes the key for you and writes the prompt. Its first line says the agent cannot publish, and that buyers see its spot changes only once you publish. The key it makes can also change a live guide's settings and price, and your profile, which buyers see at once (see When buyers see the agent's work). If you closed the card with its X, the AI agent card switch in Settings brings it back.
| Step | What it does |
|---|---|
| Create a key for your agent | Create key makes a key with Read and Write access, named after the guide, and shows it once. It is already filled into the setup below it. It stops working after 90 days. Each guide keeps one such key: when you come back, this step shows its first characters, the day it was made and the day it stops, and Replace key makes a new one and stops the old one at once, for when you lost the key or it is about to stop. Keys made here show in API keys in Settings, where you can revoke them. This key never publishes; for one that may, make it in Settings. |
| Add Sceniq to your agent | A row of chips names the apps: Claude Code, Claude Desktop, Cursor and Codex CLI. Pick yours, and the setup below switches to that app, with the same line on where it goes as in Settings. Copy it with its Copy button. |
| Tell it what to do | Two chips pick the job, Add new spots or Complete my spots, and the prompt below switches to it. Copy it with its Copy button (screen readers call it Copy the prompt), add your own material at its end and send it to your agent. |
| Job | The prompt asks the agent to |
|---|---|
| Add new spots | Add spots to this guide, each with a title, the pin from its Google Maps link, its categories (what the place is), one or two sentences of description and the guide's properties. You paste your Google Maps links and notes at its end. |
| Complete my spots | Go through this guide's spots and fill in missing categories, a description or a property, without changing what you wrote. You paste your notes, captions or old posts at its end. |
Both prompts name the guide and its id, and tell the agent to use only your material and what the links show, to leave a field empty rather than guess, and not to publish. The dialog opens on Complete my spots when some spots still miss a description or a property, and on Add new spots otherwise. The agent's spots and edits show up in your spot list as it works.
Add spots on Spots offers the same job chips and prompt on its Sceniq MCP card, without the key and setup steps: connect your agent once in Settings, then send it the prompt from there. See Import spots.
Keeping keys safe
- Give each agent or computer its own key with a clear name, so you can revoke one without stopping the others.
- Keep a key in the agent's settings, as the snippets do. Never paste it into a chat, a prompt or a file you share.
- Pick an expiry for a short job.
- Revoke a key the moment you stop using it, or if you think someone else has seen it.
- Changes made through a key are marked on the guide, and Sceniq's store review sees that mark.
When the agent is refused
An agent reports a refusal with a code. The ones about keys:
| Code | Why | What to do |
|---|---|---|
invalid_api_key | The key is mistyped, revoked or expired, the person who made it is no longer on the creator account, or the account is suspended. | Check the setup, or make a new key. While the account is suspended, no key works. |
api_scope_required | A Read only key tried to change something, or a key without Publish tried to publish. | Make a key with the access the job needs. |
api_key_not_allowed | The step stays with you (see What stays with you). | Do it in the studio. |
agreement_outdated | A new creator agreement waits for you to accept it. | Accept it in the studio; then the agent can carry on. |
rate_limited | Too many requests in a short time. | Nothing: the agent is told how long to wait. |
The developer docs
The API itself, with every operation and its fields, is in the developer docs:
- Connect an agent: the setup for each app, in more detail.
- API keys and scopes: the scopes as the API sees them.
- Build a guide with an agent: the order an agent works in.
- Rate limits: how many requests and uploads a key may make.