Module streaming 

Streaming LLM provider support for Gestura

This module provides streaming capabilities for LLM responses, enabling real-time token-by-token delivery to the frontend with cancellation support.

Modules

pricing

Pricing per 1M tokens (input/output) for various providers Prices are in USD and updated as of January 2026

Structs

AnthropicStreamRequest

Stream a response from Anthropic Claude API

CancellationToken

Cancellation token for streaming requests

PublicNarration

Structured public narration content rendered between major loop events.

TaskRuntimeSnapshot

Runtime-authored task scheduler snapshot streamed to UI surfaces.

TaskRuntimeTaskView

Compact task view for runtime-authored task-state updates.

Enums

CancellationDisposition

Requested interruption disposition for a streaming request.

NarrationStage

Public-facing narration stage for brief between-tool updates.

ShellOutputStream

Which output stream a shell chunk originated from.

ShellProcessState

Lifecycle state of a shell process.

ShellSessionState

Lifecycle state of a long-lived interactive shell session.

StreamChunk

A chunk of streaming response

TokenUsageStatus

Token usage status indicator for visual feedback

Functions

split_think_blocks

Split a complete assistant message into (user-facing text, optional thinking) based on <think>...</think> blocks.

start_streaming

Start a streaming LLM request based on config.

start_streaming_with_fallback

Start streaming with fallback to secondary provider on failure Implements jittered exponential backoff with rate-limit-aware delay selection before falling back.

stream_anthropic

stream_gemini

Stream a response from Google Gemini API (Generative Language API).

stream_ollama

Stream a response from Ollama local API

stream_openai