oracle

To bundle prompts and files for AI models, provide a task prompt and select files using --file patterns, then run the oracle CLI to bundle them into a single contextual request that preserves repository context for AI answers.

Yes, you can maintain long-running context across multiple AI code analysis runs by using session-based analysis, which stores sessions to preserve repository context and continuity across multiple invocations.

You attach files, directories, and globs to an AI prompt using --file patterns, which automatically respect ignore rules and size limits to manage file-attachment handling efficiently for contextual AI requests.

Yes, oracle works with both browser-based and API-based AI engines, allowing you to switch between engine choices and manage attachments for efficient reasoning during long-context code reviews or data extraction.

The best way to perform long-context code reviews using repository context is to bundle your selected files and prompt into a single contextual request, enabling the AI model to answer with full repository context for accurate investigations.

You configure session management and file attachments using a frontmatter-driven configuration workflow that coordinates optional scripts, references, and assets directories to manage contextual AI requests and session-backed investigations.