Settings

The Settings page is where you manage your account, credentials, connections, memory, and workspace preferences. Open it from the sidebar or press Cmd/Ctrl + ,.

Account

Sign in (or out) and view your plan here. Signing in is required before Pencel will use any model — including your own API key. On first launch you accept the Terms of Service and sign in as part of onboarding; after that it persists across launches. See Accounts & sign-in for how sign-in, subscriptions, and BYOK relate.

The Cost & Usage section shows your remaining credits and per-run spend across all runs.

Credentials

All keys and secrets Pencel uses on your behalf live in a single Credentials tab, listed in one table. Each row's Kind column tells you what it is:

• AI model — Anthropic, Google, OpenAI keys
• Tool — keys for first-party tools that need one (e.g. Tavily web search)
• Data source — credentials for source-aware connections (SAM.gov, HigherGov, GovCon, Finnhub), shared across every workflow that uses that source
• Custom — free-form named secrets that your workflows can look up by name

You only need keys for the providers and sources you plan to use.

Reading the table

Column	What it tells you
Credential	The masked key on file, the number of saved fields, or "No key"
Status	Connected, Pencel-managed (running on Pencel's shared key), Not set, or Failed
Used by	How many workflows depend on this credential
Last verified	When the key last passed a test — "Never" if you haven't run one

The needs attention count above the table only includes credentials that are actually a problem: ones that failed, and ones that are unset while workflows depend on them. A key you simply haven't set up, that nothing uses, is not counted.

Use the filter buttons to show one kind at a time, or search — search matches workflow names too, so typing a workflow name shows the credentials it needs.

Adding an API Key

1. Open Settings → Credentials from the sidebar.
2. Click the row for the provider you want to configure (Anthropic, Google, or OpenAI).
3. Paste your API key into the editor panel that opens.
4. Click Save, or Save & test to check it against the provider right away.

If the key is rejected, the row shows Failed with the reason until you replace it. Removing a key asks for confirmation first.

Where to Get API Keys

Provider	Key Source	What You Need
Anthropic	console.anthropic.com	An API key from your Anthropic account
Google	aistudio.google.com	A Gemini API key from Google AI Studio
OpenAI	platform.openai.com	An API key from your OpenAI account

INFO

You can use your company's API keys. Any valid key works, whether it's from a personal account or an organization account. Usage and billing go through the provider account that owns the key.

Security

Your API keys are stored in your operating system's secure keychain:

• macOS — Keychain Access
• Windows — Windows Credential Manager
• Linux — libsecret (GNOME Keyring / KWallet)

Keys are never stored in plain text, never written to the database, and never sent anywhere except to the respective AI provider during authenticated requests. See Data Handling for more details.

WARNING

Never share your API keys. If you suspect a key has been compromised, revoke it immediately through the provider's console and generate a new one.

Updating or Removing a Key

To update a key, enter the new key in the same field and click Save. The old key is replaced.

To remove a key, clear the field and save. Agents using that provider's models will not be able to run until a new key is provided.

Connections

The Connections section of Settings provides quick access to your external tool integrations. From here you can:

• View all configured connections and their status
• Add new connections
• Test existing connections
• Disconnect or remove connections

For detailed instructions on setting up specific integrations, see the Integrations section.

Memory

The Memory section is where you review and curate what your agents have learned. From here you can edit, disable, boost, or delete memory items, and add long-term knowledge directly. Agents draw on memory automatically as part of Knowledge when they run. See Memory for the model and Using Memory for hands-on steps.

General Settings

Workspace Preferences

• Workspace Name — A label for your workspace, shown in the title bar
• Default Model — The model used when creating new agents (you can override per agent)

TIP

If most of your agents use the same model, set it as the default here. You can always override the model for individual agents that need something different.