oracle

You can attach selected project files to your prompt using glob patterns, providing the AI model with necessary repository context for accurate code reviews and analysis. This allows the model to process your code with full file awareness.

Yes, you can preview token costs using the dry-run feature before executing a prompt with file attachments. This calculates the expected token consumption of your bundled prompt and files without making an actual API or browser engine call.

Yes, the CLI supports session management across both browser and API engines. This allows you to maintain continuous execution states and manage prompt runs seamlessly whether you are using a browser interface or an API backend.

The best way to provide repository context is bundling your prompt with selected files from your project. This file attachment pattern gives the target model direct access to relevant code, enabling richer and more accurate AI responses for analysis tasks.

You attach files using flexible glob patterns to specify which project files to bundle with your prompt. This allows you to precisely target and include relevant code files for repository analysis or code review tasks across browser or API engines.

When bundling file context for large repositories, limitations depend on the target model's token limits. Using the dry-run token cost preview helps you anticipate and manage these constraints before executing the prompt across browser or API engines.