Streams chat completions from the CodeAlive API for code-aware conversations with knowledge of your codebase. Args: messages: List of message objects with "role" and "content" fields Example: [ {"role": "system", "content": "Analyze the authentication flow"}, {"role": "user", "content": "How does the login process work?"} ] data_sources: List of data source objects to include in the context Can include workspace IDs (to chat about all repositories in the workspace) or specific repository IDs for more focused analysis. Example: [ {"type": "repository", "id": "67f664fd4c2a00698a52bb6f"}, {"type": "workspace", "id": "5e8f9a2c1d3b7e4a6c9d0f8e"} ] stream: Whether to stream the response (must be true, non-streaming is not supported) Default: true conversation_id: Optional ID to continue a previous conversation Example: "conv_6789f123a456b789c123d456" debug: Whether to include debug information in the response Default: false Returns: The generated completion text with code understanding from specified repositories/workspaces. The response will incorporate knowledge from the specified code repositories. Examples: 1. Start a new conversation about authentication using a specific repository: chat_completions( messages=[{"role": "user", "content": "Explain the authentication flow in this code"}], data_sources=[{"type": "repository", "id": "67f664fd4c2a00698a52bb6f"}] ) 2. Start a new conversation using an entire workspace: chat_completions( messages=[{"role": "user", "content": "How do the microservices communicate with each other?"}], data_sources=[{"type": "workspace", "id": "5e8f9a2c1d3b7e4a6c9d0f8e"}] ) 3. Continue an existing conversation: chat_completions( messages=[{"role": "user", "content": "How is password hashing implemented?"}], conversation_id="conv_6789f123a456b789c123d456" ) Note: - Either conversation_id OR data_sources is typically provided - When creating a new conversation, data_sources is optional if the API key has exactly one assigned data source - When continuing a conversation, conversation_id is required - The conversation maintains context across multiple messages - Messages should be in chronological order with the newest message last - Choose between workspace-level access (for broader context) or repository-level access (for targeted analysis) based on your query needs - If a user is working in a local git repository that matches one of the indexed repositories in CodeAlive (by URL), you can leverage this integration for enhanced code understanding

Gets all available data sources (repositories and workspaces) for the user's account. A data source is a code repository or workspace that has been indexed by CodeAlive and can be used for code search and chat completions. Args: alive_only: If True (default), returns only data sources in "Alive" state ready for use with chat. If False, returns all data sources regardless of processing state. Example: true Returns: A formatted list of available data sources with the following information for each: - id: Unique identifier for the data source, used in other API calls - name: Human-readable name of the repository or workspace - type: The type of data source ("Repository" or "Workspace") - url: URL of the repository (for Repository type only) - repositoryIds: List of repository IDs included in the workspace (for Workspace type only) - state: The processing state of the data source (if alive_only=false) Examples: 1. Get only ready-to-use data sources: get_data_sources() 2. Get all data sources including those still processing: get_data_sources(alive_only=false) Note: Data sources in "Alive" state are fully processed and ready for search and chat. Other states include "New" (just added), "Processing" (being indexed), "Failed" (indexing failed), etc. For repositories, the URL can be used to match with local git repositories to provide enhanced context for code understanding. For workspaces, the repositoryIds can be used to identify and work with individual repositories that make up the workspace. Use the returned data source IDs with the search_code and chat_completions functions.

Search for code snippets across the provided data sources using natural language or code patterns. Args: query: The search query - can be natural language ("find authentication code") or code patterns ("function getUserById") For best results, be specific and include relevant keywords or function/class names Example: "implement JWT token validation" data_source_ids: List of data source IDs to search in (required) Can be workspace IDs (to search across all repositories in the workspace) or individual repository IDs for more targeted searches. Example: ["67f664fd4c2a00698a52bb6f", "5e8f9a2c1d3b7e4a6c9d0f8e"] mode: Search mode (case-insensitive): - "auto": (Default) RECOMMENDED - Intelligently adapts search depth based on query complexity - "fast": Quick scan for exact matches, best for simple queries and large codebases - "fast_deeper": Balanced search with moderate semantic analysis, good for general use - "deep": Use SPARINGLY - Resource-intensive thorough semantic analysis, only for very complex conceptual queries when other modes fail to yield results Example: "auto" include_content: Whether to include the full file content in results (default: true) Set to false for faster, more concise results when only locations are needed Example: true Returns: Formatted search results including: - Source repository/workspace name and type - File path - Line numbers - Code snippet showing the matching section - Full file content (if include_content=true) Examples: 1. Find authentication implementation (using default auto mode - recommended): search_code(query="user authentication implementation", data_source_ids=["repo123"]) 2. Find a specific function quickly: search_code(query="calculateTotalPrice function", data_source_ids=["repo123"], mode="fast") 3. Search across an entire workspace: search_code(query="database connection", data_source_ids=["workspace456"]) 4. Search across specific repositories from different workspaces: search_code(query="authentication flow", data_source_ids=["repo123", "repo789"]) 5. Get concise results without full file contents: search_code(query="password reset", data_source_ids=["repo123"], include_content=false) Note: - At least one data_source_id must be provided - All data sources must be in "Alive" state - The API key must have access to the specified data sources - Always start with "auto" mode first, as it intelligently chooses the appropriate search strategy - The "deep" mode should only be used when absolutely necessary as it's resource-intensive - For finding specific implementations, include function names in your query - For understanding architectural patterns, use natural language descriptions