Skip to content

CLAUDE.md

Latest commit

 

History

History
34 lines (29 loc) · 1.64 KB

CLAUDE.md

File metadata and controls

34 lines (29 loc) · 1.64 KB

Streaming Engine Development Guide

What This Project Is

Streaming Engine is an audio processing server — it processes audio on the fly via URL parameters (like Thumbor/Imagor for images, but for audio). It is not an AI product. The MCP server integration exists to let LLMs use the audio processing API as a tool, but the core project is a deterministic audio processing pipeline built on FFmpeg.

Build & Test Commands

  • just — List available recipes
  • just dev — Run with auto-reload
  • just build — Build the project
  • just test — Run all tests
  • just test-name <name> — Run a specific test
  • just bench — Run benchmarks
  • just lint — Run linter (clippy)
  • just fmt — Format code
  • just check — Full check: format, lint, build, test

Project Structure

  • src/ — Core streaming engine (Rust)
  • crates/ffmpeg — Safe FFmpeg wrapper
  • crates/ffmpeg-sys — Raw FFI bindings to FFmpeg
  • mcp-server/ — MCP integration for LLM tool use (Node.js)
  • config/ — YAML configuration files
  • benches/ — Performance benchmarks
  • scripts/ — Deployment and CI scripts

Code Style

  • Imports: Group std, external crates, then local modules
  • Error Handling: Use color_eyre::Result, thiserror for custom errors
  • Logging: Use tracing with structured logging and #[instrument] for functions
  • Types: Prefer explicit types, use Uuid for IDs, DateTime<Utc> for timestamps
  • Naming: snake_case for functions/variables, PascalCase for types, modules in snake_case
  • Async: Use tokio::main and async/await throughout
  • API: Use Axum with State extraction and Json responses