indicia CLI vs. direct provider

You can reach the gateway from a coding tool in two ways: launch the tool through the indicia CLI, or add the gateway to the tool's own config as a provider. Both use the same endpoints and the same token types.

indicia codeCopy Anchor Link

indicia code [tool] starts Codex (the default), Claude Code or opencode with the gateway already configured. Arguments after -- go to the tool:

indicia code
indicia code claude -- --model claude-sonnet-5-5
indicia code opencode

On every launch the CLI fetches your configuration from the gateway and passes it to the tool for that run only:

• The gateway as the model provider, with your personal token.
• Your team's MCP servers, including the team kennisbank.
• Your team's instructions.
• The current model classes and their names, and the default model.

It does not write to the tool's own config files, so a launch without indicia still uses your normal setup. The CLI also updates itself from the gateway.

Direct providerCopy Anchor Link

You add the gateway to the tool's config yourself, as on the following pages. This works with any token, including a service token, and in tools the CLI does not launch. The config is yours to maintain:

• No team MCP servers or team instructions unless you add them.
• The model list is what you typed. Check GET /v1/models when models change.

Which to useCopy Anchor Link

Use indicia code for day-to-day coding. Use a direct provider for tools the CLI does not launch, for unattended jobs with a service token, or when you want the gateway in your tool's normal config.