Log error message - for serious issues that affect functionality.

Log info message - for successful operations and progress updates.

* Check if an order exists * @param id - Order ID * @returns true if order exists, false otherwise

Return [(segment_metadata, talk_text), ...] in order. Each talk's text runs from its start_anchor to the start of the next talk's anc

Write one .txt per talk plus a segments.json index. Returns the list of .txt paths written.

Log warning message - for recoverable issues and fallbacks.

Take a screenshot for debugging.

* Read an order by ID * @param id - Order ID * @returns The order if found, undefined otherwise

* Get the total count of orders * @returns Total number of orders in the store

Log debug message - for detailed troubleshooting (use sparingly).

* Load state from JSON file * If file doesn't exist or is invalid, starts with empty state * @returns Number of orders loaded

* Delete an order by ID * @param id - Order ID * @returns true if deleted, false if not found

* List all orders with optional filtering * @param filter - Optional filter criteria * @returns Array of orders matching the filter

Log success message - for completed operations.

Get video by ID

Get video by ID

Complete workflow: login and schedule a session. Args: session: Session details to schedule. Returns:

Update video fields

* Save current state to JSON file * @returns true if saved successfully, false otherwise

Update video fields

Display text with optional LLM pattern highlighting and statistics

Load categories from configured source. Returns: The categories.

Embed a category with vector store. If the category is already in the vector store, return the cached embedding. If the category is n

Check if a category exists in the vector store. Args: category_path: The category path to check. Returns: Tr

Use embedding first to get 10 candidates, then LLM to refine to 3. Args: text: The text to narrow categories based on.

Returns the full YouTube URL for the video.

Get categories. Returns: The categories.

Compute cosine similarity between embeddings. Args: embedding1: The first embedding. embedding2: The second embedding

Embed a single text. Args: text: The text to embed. Returns: The embedding of the text.

* Clear all orders (useful for testing) * @returns Number of orders cleared

Create a bar chart comparing a single metric across runs.

Find the most similar categories to a query embedding. Args: query_embedding: The embedding to search for similar categories.

Get cached embedding for a category if it exists. Args: category_path: The category path to look up. Returns:

Get a logger instance for a module. Args: name: Logger name, typically __name__ of the calling module level: Logging level (DEBUG

Check if the vector store is available.

Select the single best category from candidates using LLM. Args: text: The text to classify. candidates: The candidat

Update a specific field in a draft (for parallel content generation)

Update a specific field in a draft (for parallel content generation)

Add a new category to the vector store. Args: category: The category to add. embedding: The category's embedding.

Full classification pipeline with detailed results. Args: text: The text to classify. max_candidates: The maximum num

Narrow categories using the configured strategy. Args: text: The text for which to narrow the categories. categories: