Check if the retrieval pipeline service is available

Check if the player has won.

Chunk text respecting paragraph boundaries

Chunk text respecting paragraph boundaries

Chunk text respecting paragraph boundaries

Simple size-based chunking

Simple size-based chunking

Simple size-based chunking

Chunk conversation using fixed number of rounds Args: conversation_id: Conversation identifier test_

Chunk conversation using fixed number of rounds Args: conversation_id: Conversation identifier test_

Combine dense and sparse results for reranking. Args: dense_results: Results from dense search sparse_results

Strategy 3: Context-aware summarization considering the query

Strategy 4: Context-aware summarization with citations

Convert Anthropic message format to OpenAI format

Convert Anthropic tool format to OpenAI format

Create basic chunks using traditional chunking

Create a LegalChunk object

Create OpenAI-compatible client.

Create an item by name.

Create a new empty index.

Create sample conversation chunks for demo

Create sample test cases based on LOCOMO structure

Create a prompt for LLM summarization. Args: question: The original question answer: The final answe

Create the system prompt for the agent Returns: System prompt string

Decode generated token IDs into audio waveforms. Args: generated_ids: Tensor of generated token IDs

Detect the best backend for current platform

Drop an item.

Check if model is cached and download if needed with progress. Args: model_name: HuggingFace model name

Execute click action with smart waiting.

Implement tool-specific logic Args: params: Tool input parameters Returns: Dict

Execute text input action.

Execute navigation action.

Execute scroll action.

Execute select option action.

Execute a single workflow step. Args: step: The step to execute

Execute tab switching (simplified for single tab replay).

Execute a tool and return the result

Execute a tool and return the result with optional compression Args: tool_name: Name of the tool to execute

Execute a tool and return the result with detailed error information Returns: Tuple of (result, error_detail)

Execute a tool and return the result with detailed error information

Execute a tool and return the result Returns: Tuple of (result, error_message)

Execute a tool and return the result

Execute a tool and return the result

Execute a tool call

Execute a tool and return results

Execute tool calls and return results. Ensures that error messages from failed tool executions are properly formatted.

Execute file upload action.

Execute wait action.

Extract action data for workflow capture. Args: action: The action object from browser-use result: T

Extract approach from metadata steps. Args: metadata: Annotator metadata Returns:

Extract audio to text transcript

Extract and format Intel x86/x64 instructions.

Extract key patterns from trajectory. Args: trajectory: Execution trajectory Returns:

Extract text content from MCP result.

Extract section/chapter/article information from legal text

Extract parameters from task description for workflow. Args: task: Task description workflow: Workfl

Extract multimodal content to text first

Extract tools used from metadata. Args: metadata: Annotator metadata Returns: L

Fallback extraction method if LLM is unavailable

Fallback to DuckDuckGo if Google API not available.

Fallback LLM evaluation using direct API call when llm_evaluator module isn't available

Find which memory cards might be relevant to the question

Format boolean value with color

Format experiences for inclusion in system prompt. Args: experiences: List of experience dictionaries

Format messages based on mode - recreated each iteration for incorrect modes

Format reranked results for response. Args: reranked: List of reranked results Returns:

Format a table into structured text.

Format TODO list for display

Format TODO list for display

Format trajectory for readability. Args: trajectory: Raw trajectory data Returns:

Generate contextual description for a chunk using Anthropic's exact approach. This follows Anthropic's Contextual Retrieval

Generate contextual description for a conversation chunk. This follows Anthropic's Contextual Retrieval approach adapted for

Generate contextual enhancements for chunks. Educational Note: Following Anthropic's Contextual Retrieval approach:

Generate a simple diff preview.

Generate semantic tags for better retrieval Args: chunk: Conversation chunk Returns:

Generate semantic tags for better retrieval Args: chunk: Conversation chunk Returns:

Generate summary cards from conversation chunks using LLM extraction. Uses an LLM to analyze conversation chunks and extract

Get tool definitions with contextual search capabilities

Get a summary of current context for context-aware compression

Get detailed error information for debugging

Get document from Dify

Get document from Dify

Get entity or community from GraphRAG index

Get entity or community from GraphRAG index

Get document from local retrieval pipeline

Get document from local retrieval pipeline

Get document/node from RAPTOR index

Get document/node from RAPTOR index

Get element information from selector map. Args: index: Element index selector_map: Browser-use sele

Get enhanced tool definitions for contextual search

Get suggestions for fixing common errors

Get suggestions for fixing common errors

Get list of files to search

Get current git branch

Get current git branch

Get git status

Get or create GitHub client.

Get or create Google Calendar service.

Get main branch name

Get main branch name