Console guide

A tour of the API Platform Console: Dashboard, Search, Workbench, API keys, Usage and Settings, and what each page shows.

The Console is the API Platform's part of the Safeguard web app. It appears when Safeguard turns the API Platform on for your workspace.

Open the Console

Open the product switcher and choose API Platform (Console), or
press G, then 5.

The sidebar has Dashboard, Search, Workbench, API keys, Usage and Settings.

When your workspace is set up for the API Platform but the gateway does not serve it yet, the Dashboard, Workbench, API keys and Usage pages show a notice, The inference gateway is not connected for this workspace yet., with the reason underneath:

The inference API is not switched on for this deployment. The gateway is off for the whole Safeguard deployment, as it is during early access.
The inference API is not enabled for this account. It is off for your workspace.

When the gateway serves your workspace, the notice does not appear.

Dashboard

Your activity over the last 14 days, counted in UTC days.

Item	Shows
Requests	Requests in the period, with the amount spent underneath.
Input tokens, Output tokens	Tokens in the period.
Active API keys	Your keys that are not revoked.
Requests per day	A bar chart, one bar per day.

API keys and Open workbench at the top take you to those pages. With no activity yet, the chart says so and suggests creating a key or trying a prompt.

Search

The same Griffin AI search as in the other Safeguard products, so you can ask about your workspace without leaving the Console.

Workbench

Try a model without writing code: choose a model, write a system prompt and a prompt, set Max tokens and Temperature, and select Run. The answer streams in with its tokens, cost, latency and time to first token. Runs spend the workspace's credit. See Workbench.

API keys

The keys you own. Under Active:

Column	Shows
Name	The key's name.
Key	The key masked as its start and end, such as `sg-api-1a2b3c4d…9f8e`. The full key is shown only once, when you create it.
Created	When the key was created.
Rate limit	`Default`, or the key's own limit, such as `120/min`.
Last used	When the key last made a request, updated at most once a minute.
This month	What the key spent this month against its monthly limit, with a bar that turns red at 90%.
Status	`Active`, or `Expired` for a key created elsewhere with an expiry date that has passed.

Revoked keys are listed below, under Revoked, with their name, masked key, when they were created and revoked, and what they spent this month. They cannot be used again; they stay listed so their use can be traced.

Create key opens the form: name, monthly spend limit and requests per minute. The next screen shows the key once.
Edit changes the name and the two limits.
Revoke ends a key after you confirm. Requests with it are refused within 30 seconds, and the key moves to Revoked.

Until the API Platform is on for your workspace, Create key is disabled.

See Authentication.

Usage

What the gateway metered, by UTC day, for the period you choose: Last 7 days, Last 14 days, Last 30 days (the default) or Last 90 days.

Requests, Input tokens, Output tokens and Cost for the period.
By model and By key: requests, input and output tokens and cost for each. A key is shown by its name and masked form, Workbench runs as Console workbench, and a key that no longer exists as Deleted key. The Console never shows a key's numeric id.
Request log: the newest requests first, with Time, Model, Key (named the same way as in By key), Input, Output, Cost, Latency and Status. A failed request shows its error code as its status. Output marked est. was billed on an estimate because the provider sent no usage. Load older requests pages back through the whole log. The log holds metadata only; prompts and outputs are never shown here.
Credits and limits: Balance, Reserved (held for requests in flight), Available, and Spent this month against the workspace's monthly limit.
Gateway settings: the workspace's monthly spend limit, Store prompts and outputs, Allow retries and fallbacks and Only use models whose provider offers zero data retention. Tenant admins and super admins can change them; everyone else sees them read-only, with the note "A tenant administrator can change these."
Stored prompts and Delete stored prompts: how many prompts are stored, and, for tenant admins and super admins, a button that deletes them all after you confirm.
Retention: how long request logs and stored prompts are kept. Until Safeguard sets a period, both read "kept until deleted (no retention period is set yet)".
Credit ledger: every credit movement, with time, entry type, amount, balance after and note. Load older entries pages back.
Add credit appears for tenant admins and super admins when card top-up is available. Otherwise the page says that card top-up is not available yet and to contact Safeguard to add credit.

See Usage and analytics and Credits and invoices.

Times on these pages are shown in your browser's time zone. The server sends them in UTC.

Settings

The same workspace settings as in the other Safeguard products, such as members, organizations, notifications and appearance, as your role and your workspace's features allow.

Before the API Platform is on

If the Console is visible but the API Platform is not on for your workspace yet:

Create key is disabled;
Run in the Workbench answers that the inference endpoint is not connected yet, and no request reaches a model;
Dashboard and Usage have no metering from the gateway to show.