Skip to Content
ConfigurationOverview

Configuration overview

Configuration is where a project keeps the things its agents need but shouldn’t contain: the API key for your OpenAI account, the connection string for your Postgres database, the Salesforce credentials, the rule that says customer phone numbers must never reach a model.

Everything here is project-scoped and reusable. You set a value once, and every agent, tool and node in the project can use it — without any of them holding a copy.

The Configuration section showing its three tabs: Configurations, LLM Policies and LLM Settings

The three tabs

TabWhat it holdsTypical question it answers
ConfigurationsNamed connection details for models, embeddings, databases, voice services and external systems, plus the rules coding agents follow”Where does the API key for our Anthropic account live?”
LLM PoliciesGuardrails on what agents may send to a model and what may come back”How do I stop customer email addresses reaching the provider?”
LLM SettingsWhich model the platform uses for each of its own features”Which model writes the SQL when I use the Data Assistant?”

They’re related but independent. You can use LLM Settings without ever creating a configuration — the platform supplies models with your plan. You can write a policy without touching either.

How they fit together

Configurations                LLM Settings                LLM Policies
──────────────                ────────────                ────────────
your API keys,          →     which model runs      +     what may be sent
DB connections,               each platform               to that model, and
service credentials           feature                     what comes back

        │                            │                           │
        └────────────┬───────────────┘                           │
                     ▼                                           ▼
              an agent runs  ─────────────────────────▶  the LLM provider

A configuration is what you can connect to. An LLM setting is which of those gets used where. A policy is what’s allowed through.

Where to start

  • Setting up a project for the first time? Start with Configurations. Most projects need at least a database connection or an integration.
  • Happy with the models qRaptor provides? You can skip Configurations entirely — see Models provided by qRaptor.
  • Handling regulated or customer data? Go to LLM Policies.
  • Want a specific model for code generation, or your own account billed instead of credits? That’s LLM Settings.
  • Building an agent that works on a repository? It needs the five rule sets under Code & Execution, plus a git credential under Integrations.

Who can change what

ℹ️

Viewing is open to every project member. Creating, editing, disabling and publishing all require the Developer role or above — Developer, Admin or Owner. Viewers see the same screens with the actions removed.

Next

Security & networkingOverview