Embed content and add embeddings to the ChromaDB collection. :param chromadb: ChromaDBCollection instance to add embeddings to :param co

Embed content and add embeddings to the Qdrant collection. :param qdrant: QdrantCollection instance to add embeddings to :param content:

Embed query string find most relevant entries from the ChromaDB collection. :param chromadb: ChromaDBCollection instance to add embeddings t

Embed query string find most relevant entries from the Qdrant collection. :param qdrant: QdrantCollection instance to add embeddings to

Example of using the embedding API.

End the secure session and clear all sensitive data

End the secure session and clear all sensitive data

Estimate and print inference memory requirements for a HuggingFace model. Uses hf-mem (https://github.com/alvarobartt/hf-mem) to fetch S

Estimate the number of prompt tokens and completion tokens using character-based approximation. Args: messages: List of

Return total VRAM and component breakdown (bytes). For MoE models, we use active_params for weight memory estimation since only active ex

Have GPT-4o evaluate the student model's response to a test problem.

Show how to use MinionToolCalling to analyze Python code.

Show how to use MinionToolCalling with custom tools.

Show how to use MinionToolCalling to extract information from a document.

Show how to use MinionToolCalling to analyze files in a directory.

Export character data to JSON string. Args: character_expose: Character exposé dictionary Returns: JSON str

Export OAuth2 clients to file (without secrets).

Export API keys to a file.

Extract credentials from task or context using regex patterns. This stores credentials securely in memory for later use.

Extracts the metadata for a given query using the firecrawl API and SERPAPI utility functions Returns: tuple: (metadata_l

Extract the story title from the generated text. Args: story_text: The full story text Returns: Extracted t

Extract text from an image using OCR via Ollama. Args: path_to_file: Base64 encoded image or path to image file model_na

Extract text content from a PDF file.

Extract text from a PDF file using PyMuPDF.

Extract text from a PDF file using PyMuPDF or pypdf as fallback.

Extract text from a folder using gitingest for comprehensive code ingestion.

Extract all URLs from message content. Args: messages: List of message dictionaries Returns

Convenience function to find optimal num_tasks_per_round Args: task: Task for minions to perform context: Context for th

Format worker attempts into a readable string for the supervisor.

Format execution time in a human-readable way.

Create a ConversationTurn from a dictionary.

Generate a character exposé from the description.

Generate image using Together AI FLUX.1-schnell API

Generate a new API key.

Use a local model to generate keywords and weights for BM25 search based on user query. Args: user_query: The user's natural lan

Generate a summary report for a specific skill based on all test evaluations.

Have GPT-4o generate specific test problems for a given skill.

Generate a story using the turn-taking protocol. Args: story_idea: The basic idea for the story targ

Generate the story using the simplified Story Minion protocol

Generate metadata for the story. Args: story_data: Story data dictionary Returns: Metadata dictionary

Get the public agent card.

Fetch the public agent card.

Fetch the public agent card.

Get agent card.

Get the first available Gemma model.

Get a list of available Ollama Turbo models Returns: List[str]: List of model names

Return a list of model names available on the server.

Get list of services with available API keys

Get a cached character exposé.

Generate the user prompt for character creation.

Format character exposé for display in the UI.

Return the current conversation history

Fetch the extended agent card.

Fetch the extended agent card.

Compute final estimates for energy and average power based solely on the measured data. This estimation is based on: - The

Return models available on the server.

Get network health status from coordinator.

Get OAuth2 access token.

Generate a summary of the project in the workspace. Returns: Dictionary containing project summary information

Get configuration for a specific MCP server

Returns a formatted string listing the supported AI providers. Returns: str: A string listing major AI providers

Get task status.

Search the web for for relevant urls using SerpAPI Args: query: The query to search the web for Returns: A list of urls

Handle a new chat message.

Check server health.

Check server health.

Check server health.

Convert an image or PDF to markdown using SmolDocling. Args: image_data: Path to image file, URL, PIL Image object, image/PDF data a

Import character data from JSON string. Args: json_content: JSON string containing character data Returns:

Initialize the clients for both methods

Initialize minion instance on startup if API key is available.

Initialize minions instance on startup if possible.

Initialize the model at server startup.

Set up the secure communication channel with attestation and key exchange

Initialize Streamlit session state variables.

Initialize session state variables

Inject stored credentials into action if available. Returns the original action if no credentials are found.

Handle chat using the Interactions API. The Interactions API provides a simpler interface for chat-like interactions

Run in interactive mode.

Run in interactive mode.

Enhanced safety validation with more flexible UI interaction support.

Check if the current model is a reasoning model. Returns: bool: True if the model supports reasoning_content output

Execute a keyboard shortcut combo in an application.

This function is kept for compatibility but now just returns the endpoint. The actual server should be started separately using the OpenAI-co

Return sorted list of GPU keys.

Return sorted list of model keys.