# Wayland → WebRTC Remote Desktop Implementation Plan

## TL;DR

> **Quick Summary**: Implement a high-performance Rust backend that captures Wayland screens via PipeWire DMA-BUF, encodes to H.264 (hardware/software), and streams to WebRTC clients with 15-25ms latency target.
>
> **Deliverables**:
> - Complete Rust backend (5,000-8,000 LOC)
> - 5 major modules: capture, encoder, buffer management, WebRTC transport, signaling
> - Configuration system and CLI
> - Basic documentation and examples
>
> **Estimated Effort**: Large (4-6 weeks full-time)
> **Parallel Execution**: YES - 4 waves
> **Critical Path**: Project setup → Capture → Encoder → WebRTC integration → End-to-end

---

## Context

### Original Request
User wants to implement a Wayland to WebRTC remote desktop backend based on three comprehensive design documents (DETAILED_DESIGN_CN.md, DESIGN_CN.md, DESIGN.md).

### Design Documents Analysis
Three detailed design documents provided:
- **DETAILED_DESIGN_CN.md**: 14,000+ lines covering architecture, components, data structures, performance targets
- **DESIGN_CN.md / DESIGN.md**: Technical design with code examples and optimization strategies

**Key Requirements from Designs:**
- Zero-copy DMA-BUF pipeline for minimal latency
- Hardware encoding support (VA-API/NVENC) with software fallback (x264)
- WebRTC transport with low-latency configuration
- 15-25ms latency (LAN), <100ms (WAN)
- 30-60 FPS, up to 4K resolution
- Adaptive bitrate and damage tracking

### Current State
- **Empty Rust project** - No source code exists
- **Cargo.toml configured** with all dependencies (tokio, pipewire, webrtc-rs, x264, etc.)
- **Design complete** - Comprehensive specifications available
- **No tests or infrastructure** - Starting from scratch

---

## Work Objectives

### Core Objective
Build a production-ready remote desktop backend that captures Wayland screen content and streams it to WebRTC clients with ultra-low latency (15-25ms) using zero-copy DMA-BUF architecture.

### Concrete Deliverables
- Complete Rust implementation in `src/` directory
- 5 functional modules: capture, encoder, buffer, webrtc, signaling
- Working CLI application (`src/main.rs`)
- Configuration system (`config.toml`)
- Basic documentation (README, usage examples)

### Definition of Done
   - [x] All major modules compile and integrate
   - [x] End-to-end pipeline works: capture → encode → WebRTC → client receives
   - [x] Software encoder (x264) functional
   - [x] Hardware encoder infrastructure ready (VA-API hooks)
   - [x] `cargo build --release` succeeds
   - [x] Basic smoke test runs without crashes
   - [x] README with setup instructions

### Must Have
- PipeWire screen capture with DMA-BUF support
- Video encoding (at least x264 software encoder)
- WebRTC peer connection and media streaming
- Signaling server (WebSocket for SDP/ICE exchange)
- Zero-copy buffer management
- Error handling and logging

### Must NOT Have (Guardrails)
- **Audio capture**: Out of scope (design only mentions video)
- **Multi-user sessions**: Single session only
- **Authentication/Security**: Basic implementation only (no complex auth)
- **Hardware encoding full implementation**: Infrastructure only, placeholders for VA-API/NVENC
- **Browser client**: Backend only, assume existing WebRTC client
- **Persistent storage**: No database or file storage
- **Advanced features**: Damage tracking, adaptive bitrate (deferred to v2)

---

## Verification Strategy

### Test Decision
- **Infrastructure exists**: NO
- **User wants tests**: YES (automated verification)
- **Framework**: `criterion` (benchmarks) + simple integration tests

### Automated Verification Approach

**For Each Module**:
1. **Unit tests** for core types and error handling
2. **Integration tests** for data flow between modules
3. **Benchmarks** for performance validation (latency < target)

**No Browser Testing Required**: Use mock WebRTC behavior or simple echo test for verification.

### If TDD Enabled

Each TODO follows RED-GREEN-REFACTOR:

**Test Setup Task** (first task):
- Install test dependencies
- Create basic test infrastructure
- Example test to verify setup

**Module Tasks**:
- RED: Write failing test for feature
- GREEN: Implement minimum code to pass
- REFACTOR: Clean up while passing tests

---

## Execution Strategy

### Parallel Execution Waves

```
Wave 1 (Start Immediately):
├── Task 1: Project structure and types
└── Task 5: Configuration system

Wave 2 (After Wave 1):
├── Task 2: Capture module (PipeWire)
├── Task 3: Buffer management
└── Task 6: Basic WebRTC echo server

Wave 3 (After Wave 2):
├── Task 4: Encoder module (x264)
└── Task 7: Signaling server

Wave 4 (After Wave 3):
├── Task 8: Integration (capture → encode → WebRTC)
├── Task 9: CLI and main entry point
└── Task 10: Documentation and examples

Critical Path: 1 → 2 → 3 → 4 → 8 → 9 → 10
Parallel Speedup: ~35% faster than sequential
```

### Dependency Matrix

| Task | Depends On | Blocks | Can Parallelize With |
|------|------------|--------|---------------------|
| 1 | None | 2, 3, 5, 6 | None (foundational) |
| 5 | None | 8, 9 | 1 |
| 2 | 1, 3 | 4 | 6 |
| 3 | 1 | 2 | 6 |
| 6 | 1 | 7, 8 | 2, 3 |
| 4 | 2 | 8 | 7 |
| 7 | 1, 6 | 8 | 4 |
| 8 | 4, 7 | 9 | None |
| 9 | 8 | 10 | 10 |
| 10 | 8, 9 | None | 9 |

---

## TODOs

- [x] 1. Project Structure and Core Types

  **What to do**:
  - Create module structure: `src/capture/`, `src/encoder/`, `src/buffer/`, `src/webrtc/`, `src/signaling/`
  - Define core types: `CapturedFrame`, `EncodedFrame`, `DmaBufHandle`, `PixelFormat`, `ScreenRegion`
  - Define error types: `CaptureError`, `EncoderError`, `WebRtcError`, `SignalingError`
  - Create `src/lib.rs` with module exports
  - Create `src/error.rs` for centralized error handling

  **Must NOT do**:
  - Implement any actual capture/encoding logic (types only)
  - Add test infrastructure (Task 5)
  - Implement configuration parsing

  **Recommended Agent Profile**:
  > - **Category**: `quick` (simple type definitions)
  > - **Skills**: `[]` (no specialized skills needed)
  > - **Skills Evaluated but Omitted**: All other skills not needed for type definitions

  **Parallelization**:
  - **Can Run In Parallel**: NO (foundational)
  - **Parallel Group**: None
  - **Blocks**: Tasks 2, 3, 5, 6
  - **Blocked By**: None

  **References**:

  **Pattern References** (existing code to follow):
  - `DESIGN_CN.md:82-109` - Core type definitions
  - `DETAILED_DESIGN_CN.md:548-596` - PipeWire data structures
  - `DETAILED_DESIGN_CN.md:970-1018` - Encoder data structures

  **API/Type References** (contracts to implement against):
  - `pipewire` crate documentation for `pw::Core`, `pw::Stream`, `pw::buffer::Buffer`
  - `webrtc` crate for `RTCPeerConnection`, `RTCVideoTrack`
  - `async-trait` for `VideoEncoder` trait definition

  **Test References**:
  - `thiserror` crate for error derive patterns
  - Standard Rust project layout conventions

  **Documentation References**:
  - `DESIGN_CN.md:46-209` - Component breakdown and data structures
  - `Cargo.toml` - Dependency versions to use

  **External References**:
  - pipewire-rs examples: https://gitlab.freedesktop.org/pipewire/pipewire-rs
  - webrtc-rs examples: https://github.com/webrtc-rs/webrtc

  **WHY Each Reference Matters**:
  - Design docs provide exact type definitions - use them verbatim
  - External examples show idiomatic usage patterns for complex crates

  **Acceptance Criteria**:

  **Automated Verification**:
  ```bash
  # Agent runs:
  cargo check
  # Assert: Exit code 0, no warnings

  cargo clippy -- -D warnings
  # Assert: Exit code 0

  cargo doc --no-deps --document-private-items
  # Assert: Docs generated successfully
  ```

  **Evidence to Capture**:
     - [x] Module structure verified: `ls -la src/`
     - [x] Type compilation output from `cargo check`
     - [x] Generated documentation files

  **Commit**: NO (group with Task 5)

---

- [x] 2. Capture Module (PipeWire Integration)

  **What to do**:
  - Implement `src/capture/mod.rs` with PipeWire client
  - Create `PipewireCore` struct: manage PipeWire main loop and context
  - Create `PipewireStream` struct: handle video stream and buffer dequeue
  - Implement frame extraction: Extract DMA-BUF FD, size, stride from buffer
  - Create async channel: Send `CapturedFrame` to encoder pipeline
  - Implement `DamageTracker` (basic version): Track changed screen regions
  - Handle PipeWire events: `param_changed`, `process` callbacks

  **Must NOT do**:
  - Implement xdg-desktop-portal integration (defer to v2)
  - Implement hardware-specific optimizations
  - Add complex damage tracking algorithms (use simple block comparison)

  **Recommended Agent Profile**:
  > - **Category**: `unspecified-high` (complex async FFI integration)
  > - **Skills**: `[]`
  > - **Skills Evaluated but Omitted**: Not applicable

  **Parallelization**:
  - **Can Run In Parallel**: YES (with Tasks 3, 6)
  - **Parallel Group**: Wave 2 (with Tasks 3, 6)
  - **Blocks**: Task 4
  - **Blocked By**: Tasks 1, 3

  **References**:

  **Pattern References**:
  - `DESIGN_CN.md:367-516` - Complete capture module implementation
  - `DETAILED_DESIGN_CN.md:542-724` - PipeWire client and stream handling
  - `DETAILED_DESIGN_CN.md:727-959` - Damage tracker implementation

  **API/Type References**:
  - `pipewire` crate: `pw::MainLoop`, `pw::Context`, `pw::Core`, `pw::stream::Stream`
  - `pipewire::properties!` macro for stream properties
  - `pipewire::spa::param::format::Format` for video format
  - `async_channel::Sender/Receiver` for async frame passing

  **Test References**:
  - PipeWire examples in pipewire-rs repository
  - DMA-BUF handling patterns in other screen capture projects

  **Documentation References**:
  - `DESIGN_CN.md:70-110` - Capture manager responsibilities
  - `DESIGN_CN.md:213-244` - Data flow from Wayland to capture

  **External References**:
  - PipeWire protocol docs: https://docs.pipewire.org/
  - DMA-BUF kernel docs: https://www.kernel.org/doc/html/latest/driver-api/dma-buf.html

  **WHY Each Reference Matters**:
  - PipeWire FFI is complex - follow proven patterns from examples
  - DMA-BUF handling requires precise memory management - reference docs for safety

  **Acceptance Criteria**:

  **Automated Verification**:
  ```bash
  # Agent runs:
  cargo check
  # Assert: No compilation errors

  # Create simple capture test
  cargo test capture::tests::test_stream_creation
  # Assert: Test passes (mock PipeWire or skip if no Wayland session)

  # Verify module compiles
  cargo build --release --lib
  # Assert: capture module in release binary
  ```

  **Evidence to Capture**:
     - [x] Module compilation output
     - [x] Test execution results
     - [x] Binary size after compilation

  **Commit**: NO (group with Task 3)

---

- [x] 3. Buffer Management Module

  **What to do**:
  - Implement `src/buffer/mod.rs` with zero-copy buffer pools
  - Create `DmaBufPool`: Manage DMA-BUF file descriptors with reuse
  - Create `EncodedBufferPool`: Manage `Bytes` for encoded frames
  - Implement `FrameBufferPool`: Unified interface for both pool types
  - Use RAII pattern: `Drop` trait for automatic cleanup
  - Implement `DmaBufHandle`: Safe wrapper around raw file descriptor
  - Add memory tracking: Track buffer lifetimes and prevent leaks

  **Must NOT do**:
  - Implement GPU memory pools (defer to hardware encoding)
  - Add complex memory allocation strategies (use simple VecDeque pools)
  - Implement shared memory (defer to v2)

  **Recommended Agent Profile**:
  > - **Category**: `unspecified-high` (unsafe FFI, memory management)
  > - **Skills**: `[]`
  > - **Skills Evaluated but Omitted**: Not applicable

  **Parallelization**:
  - **Can Run In Parallel**: YES (with Tasks 2, 6)
  - **Parallel Group**: Wave 2 (with Tasks 2, 6)
  - **Blocks**: Task 2
  - **Blocked By**: Task 1

  **References**:

  **Pattern References**:
  - `DESIGN_CN.md:518-617` - Frame buffer pool implementation
  - `DETAILED_DESIGN_CN.md:287-299` - Buffer module design
  - `DESIGN_CN.md:1066-1144` - Buffer sharing mechanisms

  **API/Type References**:
  - `std::collections::VecDeque` for buffer pools
  - `std::os::unix::io::RawFd` for file descriptors
  - `bytes::Bytes` for reference-counted buffers
  - `std::mem::ManuallyDrop` for custom Drop logic

  **Test References**:
  - Rust unsafe patterns for FFI
  - RAII examples in Rust ecosystem

  **Documentation References**:
  - `DESIGN_CN.md:182-209` - Buffer manager responsibilities
  - `DESIGN_CN.md:1009-1064` - Zero-copy pipeline stages

  **External References**:
  - DMA-BUF documentation: https://www.kernel.org/doc/html/latest/driver-api/dma-buf.html
  - `bytes` crate docs: https://docs.rs/bytes/

  **WHY Each Reference Matters**:
  - Unsafe FFI requires precise patterns - RAII prevents resource leaks
  - Reference design shows proven zero-copy architecture

  **Acceptance Criteria**:

  **Automated Verification**:
  ```bash
  # Agent runs:
  cargo test buffer::tests::test_dma_buf_pool
  # Assert: Pool allocates and reuses buffers correctly

  cargo test buffer::tests::test_encoded_buffer_pool
  # Assert: Bytes pool works with reference counting

  cargo test buffer::tests::test_memory_tracking
  # Assert: Memory tracker detects leaks (if implemented)
  ```

  **Evidence to Capture**:
     - [x] Test execution results
     - [x] Memory usage check (valgrind or similar if available)
     - [x] Pool performance metrics

  **Commit**: YES
  - Message: `feat(buffer): implement zero-copy buffer management`
  - Files: `src/buffer/mod.rs`, `src/lib.rs`
  - Pre-commit: `cargo test --lib`

---

- [x] 4. Encoder Module (Software - x264)

  **What to do**:
  - Implement `src/encoder/mod.rs` with encoder trait
  - Define `VideoEncoder` trait with `encode()`, `reconfigure()`, `request_keyframe()`
  - Create `X264Encoder` struct: Wrap x264 software encoder
  - Implement encoder initialization: Set low-latency parameters (ultrafast preset, zerolatency tune)
  - Implement frame encoding: Convert DMA-BUF to YUV, encode to H.264
  - Use zero-copy: Map DMA-BUF once, encode from mapped memory
  - Output encoded data: Wrap in `Bytes` for zero-copy to WebRTC
  - Implement bitrate control: Basic CBR or VBR

  **Must NOT do**:
  - Implement VA-API or NVENC encoders (defer to v2, just add trait infrastructure)
  - Implement adaptive bitrate control (use fixed bitrate)
  - Implement damage-aware encoding (encode full frames)

  **Recommended Agent Profile**:
  > - **Category**: `unspecified-high` (video encoding, low-latency optimization)
  > - **Skills**: `[]`
  > - **Skills Evaluated but Omitted**: Not applicable

  **Parallelization**:
  - **Can Run In Parallel**: NO
  - **Parallel Group**: Wave 3 (only)
  - **Blocks**: Task 8
  - **Blocked By**: Task 2

  **References**:

  **Pattern References**:
  - `DESIGN_CN.md:620-783` - Complete encoder module implementation
  - `DESIGN_CN.md:1249-1453` - Low-latency encoder configuration
  - `DETAILED_DESIGN_CN.md:963-1184` - Video encoder trait and implementations

  **API/Type References**:
  - `x264` crate: `x264::Encoder`, `x264::Params`, `x264::Picture`
  - `async-trait` for `#[async_trait] VideoEncoder`
  - `bytes::Bytes` for zero-copy output
  - `async_trait::async_trait` macro

  **Test References**:
  - x264-rs examples: https://github.com/DaGenix/rust-x264
  - Low-latency encoding patterns in OBS Studio code

  **Documentation References**:
  - `DESIGN_CN.md:112-148` - Encoder pipeline responsibilities
  - `DESIGN_CN.md:248-332` - Technology stack and encoder options
  - `DESIGN_CN.md:1376-1411` - x264 low-latency parameters

  **External References**:
  - x264 documentation: https://code.videolan.org/videolan/x264/
  - H.264 codec specification

  **WHY Each Reference Matters**:
  - Low-latency encoding requires precise parameter tuning - use documented presets
  - x264 API is complex - examples show correct usage

  **Acceptance Criteria**:

  **Automated Verification**:
  ```bash
  # Agent runs:
  cargo test encoder::tests::test_x264_init
  # Assert: Encoder initializes with correct parameters

  cargo test encoder::tests::test_encode_frame
  # Assert: Frame encodes successfully, output is valid H.264

  # Verify encoding performance
  cargo test encoder::tests::benchmark_encode --release
  # Assert: Encoding latency < 20ms for 1080p frame
  ```

  **Evidence to Capture**:
     - [x] Test execution results
     - [x] Encoding latency measurements
     - [x] Output bitstream validation (using `ffprobe` if available)

  **Commit**: YES
  - Message: `feat(encoder): implement x264 software encoder`
  - Files: `src/encoder/mod.rs`, `src/lib.rs`
  - Pre-commit: `cargo test encoder`

---

- [x] 5. Configuration System and Test Infrastructure

  **What to do**:
  - Create `config.toml` template: Capture settings, encoder config, WebRTC config
  - Implement `src/config.rs`: Parse TOML with `serde`
  - Define config structs: `CaptureConfig`, `EncoderConfig`, `WebRtcConfig`
  - Add validation: Check reasonable value ranges, provide defaults
  - Create CLI argument parsing: Use `clap` for command-line overrides
  - Set up test infrastructure: Add test dependencies to Cargo.toml
  - Create integration test template: `tests/integration_test.rs`
  - Set up benchmarking: Add `criterion` for latency measurements

  **Must NOT do**:
  - Implement hot reload of config
  - Add complex validation rules (basic range checks only)
  - Implement configuration file watching

  **Recommended Agent Profile**:
  > - **Category**: `quick` (simple config parsing, boilerplate)
  > - **Skills**: `[]`
  > - **Skills Evaluated but Omitted**: Not applicable

  **Parallelization**:
  - **Can Run In Parallel**: YES (with Task 1)
  - **Parallel Group**: Wave 1 (with Task 1)
  - **Blocks**: Tasks 8, 9
  - **Blocked By**: None

  **References**:

  **Pattern References**:
  - `DESIGN_CN.md:90-95` - Capture config structure
  - `DESIGN_CN.md:124-130` - Encoder config structure
  - `DESIGN_CN.md:169-180` - WebRTC config structure

  **API/Type References**:
  - `serde` derive macros: `#[derive(Serialize, Deserialize)]`
  - `toml` crate: `from_str()` for parsing
  - `clap` crate: `Parser` trait for CLI

  **Test References**:
  - `criterion` examples: https://bheisler.github.io/criterion.rs/
  - Integration testing patterns in Rust

  **Documentation References**:
  - `DESIGN_CN.md:248-259` - Dependencies including config tools
  - Configuration file best practices

  **External References**:
  - TOML spec: https://toml.io/
  - clap documentation: https://docs.rs/clap/

  **WHY Each Reference Matters**:
  - Config structure defined in designs - implement exactly
  - Standard Rust patterns for config parsing

  **Acceptance Criteria**:

  **Automated Verification**:
  ```bash
  # Agent runs:
  cargo test config::tests::test_parse_valid_config
  # Assert: Config file parses correctly

  cargo test config::tests::test_cli_overrides
  # Assert: CLI args override config file

  cargo test --all-targets
  # Assert: All tests pass (including integration template)

  cargo bench --no-run
  # Assert: Benchmarks compile successfully
  ```

  **Evidence to Capture**:
     - [x] Config parsing test results
     - [x] Test suite execution output
     - [x] Benchmark compilation success

  **Commit**: YES (grouped with Task 1)
  - Message: `feat: add project structure, types, and config system`
  - Files: `src/lib.rs`, `src/error.rs`, `src/config.rs`, `config.toml`, `Cargo.toml`, `tests/integration_test.rs`, `benches/`
  - Pre-commit: `cargo test --all`

---

- [x] 6. WebRTC Transport Module

  **What to do**:
  - Implement `src/webrtc/mod.rs` with WebRTC peer connection management
  - Create `WebRtcServer` struct: Manage `RTCPeerConnection` instances
  - Create `PeerConnection` wrapper: Encapsulate `webrtc` crate types
  - Implement video track: `TrackLocalStaticSample` for encoded frames
  - Implement SDP handling: `create_offer()`, `set_remote_description()`, `create_answer()`
  - Implement ICE handling: ICE candidate callbacks, STUN/TURN support
  - Configure low-latency: Minimize playout delay, disable FEC
  - Implement data channels: For input events (mouse/keyboard)

  **Must NOT do**:
  - Implement custom WebRTC stack (use webrtc-rs as-is)
  - Implement TURN server (configure external servers)
  - Implement complex ICE strategies (use default)

  **Recommended Agent Profile**:
  > - **Category**: `unspecified-high` (WebRTC protocol, async networking)
  > - **Skills**: `[]`
  > - **Skills Evaluated but Omitted**: Not applicable

  **Parallelization**:
  - **Can Run In Parallel**: YES (with Tasks 2, 3)
  - **Parallel Group**: Wave 2 (with Tasks 2, 3)
  - **Blocks**: Task 7, 8
  - **Blocked By**: Task 1

  **References**:

  **Pattern References**:
  - `DESIGN_CN.md:786-951` - Complete WebRTC module implementation
  - `DESIGN_CN.md:1573-1738` - Low-latency WebRTC configuration
  - `DETAILED_DESIGN_CN.md:270-286` - WebRTC transport module design

  **API/Type References**:
  - `webrtc` crate: `RTCPeerConnection`, `RTCVideoTrack`, `RTCDataChannel`
  - `webrtc::api::APIBuilder` for API initialization
  - `webrtc::peer_connection::sdp` for SDP handling
  - `webrtc::media::Sample` for video samples

  **Test References**:
  - webrtc-rs examples: https://github.com/webrtc-rs/webrtc/tree/main/examples
  - WebRTC protocol specs: https://www.w3.org/TR/webrtc/

  **Documentation References**:
  - `DESIGN_CN.md:150-181` - WebRTC transport responsibilities
  - `DESIGN_CN.md:348-360` - WebRTC library options
  - `DESIGN_CN.md:1577-1653` - Low-latency WebRTC configuration

  **External References**:
  - WebRTC MDN: https://developer.mozilla.org/en-US/docs/Web/API/WebRTC_API
  - ICE specification: https://tools.ietf.org/html/rfc8445

  **WHY Each Reference Matters**:
  - WebRTC is complex protocol - use proven library and follow examples
  - Low-latency config requires precise parameter tuning

  **Acceptance Criteria**:

  **Automated Verification**:
  ```bash
  # Agent runs:
  cargo test webrtc::tests::test_peer_connection_creation
  # Assert: Peer connection initializes with correct config

  cargo test webrtc::tests::test_sdp_exchange
  # Assert: Offer/Answer exchange works correctly

  cargo test webrtc::tests::test_video_track
  # Assert: Video track accepts and queues samples
  ```

  **Evidence to Capture**:
     - [x] Test execution results
     - [x] SDP output (captured in test logs)
     - [x] ICE candidate logs

  **Commit**: YES
  - Message: `feat(webrtc): implement WebRTC transport with low-latency config`
  - Files: `src/webrtc/mod.rs`, `src/lib.rs`
  - Pre-commit: `cargo test webrtc`

---

- [x] 7. Signaling Server

  **What to do**:
  - Implement `src/signaling/mod.rs` with WebSocket signaling
  - Create `SignalingServer` struct: Manage WebSocket connections
  - Implement session management: Map session IDs to peer connections
  - Implement SDP exchange: `send_offer()`, `receive_answer()`
  - Implement ICE candidate relay: `send_ice_candidate()`, `receive_ice_candidate()`
  - Handle client connections: Accept, authenticate (basic), track sessions
  - Use async IO: `tokio-tungstenite` or `tokio` WebSocket support

  **Must NOT do**:
  - Implement authentication/authorization (allow all connections)
  - Implement persistent storage (in-memory sessions only)
  - Implement NAT traversal beyond ICE (no STUN/TURN server hosting)

  **Recommended Agent Profile**:
  > - **Category**: `unspecified-low` (simple WebSocket server)
  > - **Skills**: `[]`
  > - **Skills Evaluated but Omitted**: Not applicable

  **Parallelization**:
  - **Can Run In Parallel**: YES (with Task 4)
  - **Parallel Group**: Wave 3 (with Task 4)
  - **Blocks**: Task 8
  - **Blocked By**: Tasks 1, 6

  **References**:

  **Pattern References**:
  - `DESIGN_CN.md:954-1007` - IPC/signaling implementation example
  - `DETAILED_DESIGN_CN.md:301-314` - Signaling module design
  - WebSocket echo server examples

  **API/Type References**:
  - `tokio::net::TcpListener` for TCP listening
  - `tokio_tungstenite` crate: `WebSocketStream`, `accept_async()`
  - `serde_json` for message serialization
  - `async_channel` or `tokio::sync` for coordination

  **Test References**:
  - WebSocket examples in tokio ecosystem
  - Signaling server patterns in WebRTC tutorials

  **Documentation References**:
  - `DESIGN_CN.md:27-34` - Signaling server in architecture
  - Session management best practices

  **External References**:
  - WebSocket protocol: https://tools.ietf.org/html/rfc6455
  - Signaling patterns: https://webrtc.org/getting-started/signaling

  **WHY Each Reference Matters**:
  - WebSocket signaling is standard WebRTC pattern - follow proven implementation
  - Session management required for multi-client support

  **Acceptance Criteria**:

  **Automated Verification**:
  ```bash
  # Agent runs:
  cargo test signaling::tests::test_websocket_connection
  # Assert: Client can connect and disconnect

  cargo test signaling::tests::test_sdp_exchange
  # Assert: SDP offer/answer relay works

  cargo test signaling::tests::test_ice_candidate_relay
  # Assert: ICE candidates forwarded correctly
  ```

  **Evidence to Capture**:
     - [x] Test execution results
     - [x] WebSocket message logs
     - [x] Session tracking verification

  **Commit**: YES
  - Message: `feat(signaling): implement WebSocket signaling server`
  - Files: `src/signaling/mod.rs`, `src/lib.rs`
  - Pre-commit: `cargo test signaling`

---

- [x] 8. End-to-End Integration

  **What to do**:
  - Implement `src/main.rs` with application entry point
  - Create pipeline orchestration: Capture → Buffer → Encoder → WebRTC
  - Integrate all modules: Connect channels and data flow
  - Implement graceful shutdown: Handle Ctrl+C, clean up resources
  - Add metrics collection: Track latency, frame rate, bitrate
  - Implement error recovery: Restart failed modules, log errors
  - Test with mock WebRTC client: Verify end-to-end flow

  **Must NOT do**:
  - Implement production deployment (local testing only)
  - Add monitoring/alerting beyond logging
  - Implement auto-scaling or load balancing

  **Recommended Agent Profile**:
  > - **Category**: `unspecified-high` (complex orchestration, async coordination)
  > - **Skills**: `[]`
  > - **Skills Evaluated but Omitted**: Not applicable

  **Parallelization**:
  - **Can Run In Parallel**: NO
  - **Parallel Group**: Wave 4
  - **Blocks**: Task 9
  - **Blocked By**: Tasks 4, 7

  **References**:

  **Pattern References**:
  - `DESIGN_CN.md:1044-1064` - Memory ownership transfer through pipeline
  - `DESIGN_CN.md:211-244` - Complete data flow
  - `DETAILED_DESIGN_CN.md:417-533` - Frame processing sequence

  **API/Type References**:
  - `tokio` runtime: `tokio::runtime::Runtime`, `tokio::select!`
  - `async_channel` for inter-module communication
  - `tracing` for structured logging

  **Test References**:
  - Integration test patterns
  - Graceful shutdown examples in async Rust

  **Documentation References**:
  - `DESIGN_CN.md:1009-1044` - Zero-copy pipeline stages
  - Error handling patterns in async Rust

  **External References**:
  - Tokio orchestration examples: https://tokio.rs/
  - Structured logging: https://docs.rs/tracing/

  **WHY Each Reference Matters**:
  - End-to-end integration requires precise async coordination
  - Zero-copy pipeline depends on correct ownership transfer

  **Acceptance Criteria**:

  **Automated Verification**:
  ```bash
  # Agent runs:
  cargo build --release
  # Assert: Binary builds successfully

  # Run with test config
  timeout 30 cargo run --release -- --config config.toml
  # Assert: Application starts, no crashes, logs show pipeline active

  # Verify metrics collection
  cargo test integration::tests::test_end_to_end_flow
  # Assert: Frame flows through complete pipeline, metrics collected
  ```

  **Evidence to Capture**:
     - [x] Application startup logs
     - [x] Pipeline flow verification logs
     - [x] Metrics output (latency, frame rate, bitrate)
     - [x] Graceful shutdown logs

  **Commit**: YES
  - Message: `feat: implement end-to-end pipeline integration`
  - Files: `src/main.rs`, `src/lib.rs`
  - Pre-commit: `cargo test integration`

---

- [x] 9. CLI and User Interface

  **What to do**:
  - Complete `src/main.rs` CLI implementation
  - Implement subcommands: `start`, `stop`, `status`, `config`
  - Add useful flags: `--verbose`, `--log-level`, `--port`
  - Implement signal handling: Handle SIGINT, SIGTERM for graceful shutdown
  - Add configuration validation: Warn on invalid settings at startup
  - Implement status command: Show running sessions, metrics
  - Create man page or help text: Document all options

  **Must NOT do**:
  - Implement TUI or GUI (CLI only)
  - Add interactive configuration prompts
  - Implement daemon mode (run in foreground)

  **Recommended Agent Profile**:
  > - **Category**: `quick` (CLI boilerplate, argument parsing)
  > - **Skills**: `[]`
  > - **Skills Evaluated but Omitted**: Not applicable

  **Parallelization**:
  - **Can Run In Parallel**: YES (with Task 10)
  - **Parallel Group**: Wave 4
  - **Blocks**: None
  - **Blocked By**: Task 8

  **References**:

  **Pattern References**:
  - CLI examples in Cargo.toml (bin section)
  - `clap` crate examples and documentation
  - Signal handling in async Rust

  **API/Type References**:
  - `clap` crate: `Parser`, `Subcommand` derives
  - `tokio::signal` for signal handling
  - `tracing` for log levels

  **Test References**:
  - clap documentation for all argument types
  - Signal handling patterns

  **Documentation References**:
  - CLI best practices: https://clig.dev/
  - `DESIGN_CN.md` - Configuration options to expose

  **External References**:
  - clap documentation: https://docs.rs/clap/

  **WHY Each Reference Matters**:
  - Good CLI design requires following established patterns
  - Signal handling critical for graceful shutdown

  **Acceptance Criteria**:

  **Automated Verification**:
  ```bash
  # Agent runs:
  cargo run --release -- --help
  # Assert: Help text shows all subcommands and flags

  cargo run --release -- start --config config.toml
  # Assert: Application starts with correct config

  cargo run --release -- status
  # Assert: Status command prints session info (or "no sessions")

  # Test signal handling
  timeout 5 cargo run --release -- start &
  PID=$!
  sleep 1
  kill -INT $PID
  wait $PID
  # Assert: Exit code 0 (graceful shutdown)
  ```

  **Evidence to Capture**:
     - [x] Help output
     - [x] Status command output
     - [x] Signal handling test results
     - [x] Error handling for invalid flags

  **Commit**: YES
  - Message: `feat(cli): implement complete CLI with subcommands`
  - Files: `src/main.rs`
  - Pre-commit: `cargo clippy`

---

- [x] 10. Documentation and Examples

  **What to do**:
  - Create `README.md`: Project overview, features, installation, usage
  - Document configuration: Explain all config options in `config.toml.template`
  - Add example usage: Show how to start server, connect client
  - Document architecture: Explain module design and data flow
  - Add troubleshooting section: Common issues and solutions
  - Create `examples/` directory: Simple client examples if needed
  - Document dependencies: List system-level dependencies (PipeWire, Wayland)
  - Add performance notes: Expected latency, resource usage

  **Must NOT do**:
  - Write extensive API documentation (use Rustdoc comments instead)
  - Create video tutorials or complex guides
  - Write marketing content (keep technical)

  **Recommended Agent Profile**:
  > - **Category**: `writing` (documentation creation)
  > - **Skills**: `[]`
  > - **Skills Evaluated but Omitted**: Not applicable

  **Parallelization**:
  - **Can Run In Parallel**: YES (with Task 9)
  - **Parallel Group**: Wave 4
  - **Blocks**: None
  - **Blocked By**: Task 8

  **References**:

  **Pattern References**:
  - Rust project README conventions
  - Existing design documents (DETAILED_DESIGN_CN.md, etc.)
  - Configuration file comments

  **API/Type References**:
  - Rustdoc: `///` and `//!` documentation comments

  **Test References**:
  - README examples in popular Rust projects
  - Documentation best practices

  **Documentation References**:
  - `DESIGN_CN.md` - Use architecture diagrams for overview
  - `Cargo.toml` - Extract dependency requirements
  - Design docs for feature descriptions

  **External References**:
  - README guidelines: https://www.makeareadme.com/
  - Rust API guidelines: https://rust-lang.github.io/api-guidelines/

  **WHY Each Reference Matters**:
  - Good documentation critical for open-source adoption
  - README first thing users see

  **Acceptance Criteria**:

  **Automated Verification**:
  ```bash
  # Agent runs:
  ls -la README.md config.toml.template
  # Assert: Files exist and are non-empty

  grep -q "Installation" README.md
  grep -q "Usage" README.md
  grep -q "Architecture" README.md
  # Assert: Key sections present

  head -20 config.toml.template
  # Assert: Template has comments explaining each option

  # Verify all public items have docs
  cargo doc --no-deps
  ls target/doc/wl_webrtc/
  # Assert: Documentation generated successfully
  ```

  **Evidence to Capture**:
     - [x] README content preview
     - [x] Config template preview
     - [x] Generated documentation listing

  **Commit**: YES
  - Message: `docs: add README, config template, and documentation`
  - Files: `README.md`, `config.toml.template`, `examples/`
  - Pre-commit: None

---

## Commit Strategy

| After Task | Message | Files | Verification |
|------------|---------|-------|--------------|
| 1, 5 | `feat: add project structure, types, and config system` | `src/`, `config.toml`, `Cargo.toml`, `tests/`, `benches/` | `cargo test --all` |
| 3 | `feat(buffer): implement zero-copy buffer management` | `src/buffer/mod.rs`, `src/lib.rs` | `cargo test --lib` |
| 2 | `feat(capture): implement PipeWire screen capture` | `src/capture/mod.rs`, `src/lib.rs` | `cargo test capture` |
| 4 | `feat(encoder): implement x264 software encoder` | `src/encoder/mod.rs`, `src/lib.rs` | `cargo test encoder` |
| 6 | `feat(webrtc): implement WebRTC transport with low-latency config` | `src/webrtc/mod.rs`, `src/lib.rs` | `cargo test webrtc` |
| 7 | `feat(signaling): implement WebSocket signaling server` | `src/signaling/mod.rs`, `src/lib.rs` | `cargo test signaling` |
| 8 | `feat: implement end-to-end pipeline integration` | `src/main.rs`, `src/lib.rs` | `cargo test integration` |
| 9 | `feat(cli): implement complete CLI with subcommands` | `src/main.rs` | `cargo clippy` |
| 10 | `docs: add README, config template, and documentation` | `README.md`, `config.toml.template`, `examples/` | None |

---

## Success Criteria

### Verification Commands
```bash
# Build and test everything
cargo build --release && cargo test --all

# Run basic smoke test
timeout 30 cargo run --release -- start --config config.toml

# Check documentation
cargo doc --no-deps

# Verify CLI
cargo run --release -- --help
```

### Final Checklist
   - [x] All 10 tasks completed
   - [x] `cargo build --release` succeeds
   - [x] `cargo test --all` passes all tests
   - [x] End-to-end pipeline verified (capture → encode → send)
   - [x] CLI fully functional with all subcommands
   - [x] README complete with installation/usage instructions
   - [x] Code compiles without warnings (`cargo clippy`)
   - [x] Documentation generated successfully
   - [x] Config template provided with comments

### Performance Validation (Optional)
   - [x] Encoding latency < 20ms for 1080p (measured via benchmarks)
   - [x] Capture latency < 5ms (measured via logging)
   - [x] Memory usage < 500MB (measured via `ps` or similar)

---

## Appendix

### Notes on Scope Boundaries

**IN SCOPE (This Implementation)**:
- Complete Rust backend implementation
- PipeWire screen capture with DMA-BUF
- x264 software encoder (production-ready)
- WebRTC transport with webrtc-rs
- WebSocket signaling server
- Basic configuration and CLI
- Zero-copy buffer management
- Basic logging and error handling

**OUT OF SCOPE (Future Work)**:
- Hardware encoder implementation (VA-API, NVENC)
- Advanced features: damage tracking, adaptive bitrate, partial region encoding
- Authentication/authorization
- Audio capture and streaming
- Multi-user session management
- Production deployment (Docker, systemd, etc.)
- Browser client implementation
- Comprehensive testing suite (unit + integration + e2e)
- Monitoring/metrics beyond basic logging

### Assumptions Made

1. **Wayland environment available**: Implementation assumes Linux with PipeWire and Wayland compositor
2. **x264 library installed**: System-level x264 library required (via pkg-config)
3. **Single-session focus**: Only one capture session at a time for simplicity
4. **Local network**: Low-latency targets assume LAN environment
5. **Existing WebRTC client**: Backend only, no browser client implementation needed
6. **No authentication**: Allow all WebSocket connections for MVP
7. **Single-threaded encoding**: No parallel encoder pipelines for MVP

### Risks and Mitigations

| Risk | Impact | Mitigation |
|------|--------|------------|
| PipeWire FFI complexity | High | Use proven patterns from pipewire-rs examples |
| DMA-BUF safety | High | Strict RAII, unsafe blocks well-documented, extensive testing |
| WebRTC integration complexity | Medium | Use webrtc-rs as-is, avoid custom implementation |
| Performance targets unmet | Medium | Benchmarking in Task 4, iterative tuning |
| Missing dependencies | Low | Clear documentation of system requirements |
| Testing challenges (requires Wayland) | Medium | Use mock objects where possible, optional tests |

### Alternatives Considered

**WebRTC Library**:
- **Chosen**: `webrtc` (webrtc-rs) - Pure Rust, active development
- **Alternative**: `datachannel` - Another pure Rust option, less mature

**Async Runtime**:
- **Chosen**: `tokio` - Industry standard, excellent ecosystem
- **Alternative**: `async-std` - More modern, smaller ecosystem

**Software Encoder**:
- **Chosen**: `x264` - Ubiquitous, mature, good quality
- **Alternative**: `openh264` - Cisco's implementation, slightly lower quality

### Dependencies Rationale

**Core**:
- `tokio`: Async runtime, chosen for ecosystem and performance
- `pipewire`: Required for screen capture
- `webrtc`: WebRTC implementation, chosen for zero-copy support
- `x264`: Software encoder fallback, ubiquitous support

**Supporting**:
- `bytes`: Zero-copy buffers, critical for performance
- `async-channel`: Async channels, simpler than tokio channels
- `tracing`: Structured logging, modern and flexible
- `serde/toml`: Configuration parsing, standard ecosystem
- `clap`: CLI parsing, excellent help generation
- `anyhow/thiserror`: Error handling, idiomatic Rust

### Known Limitations

1. **Linux-only**: Wayland/PipeWire specific to Linux
2. **Requires Wayland session**: Cannot run in headless or X11 environments
3. **Hardware encoding deferred**: Only x264 in v1
4. **No audio**: Video-only in v1
5. **Basic signaling**: No authentication, persistence, or advanced features
6. **Single session**: Only one capture session at a time
7. **Local testing**: No cloud deployment guidance
8. **Minimal testing**: Basic integration tests, no comprehensive test suite

### Testing Environment Requirements

To run tests and smoke tests, you need:
- Linux distribution with Wayland
- PipeWire installed and running
- x264 development libraries (`libx264-dev` on Ubuntu/Debian)
- Rust toolchain (stable)
- Optional: Wayland compositor for full integration testing

### Performance Baseline

Expected performance (based on design docs):
- **Capture latency**: 1-2ms (DMA-BUF from PipeWire)
- **Encoding latency**: 15-25ms (x264 ultrafast)
- **WebRTC overhead**: 2-3ms (RTP packetization)
- **Total pipeline**: 18-30ms (excluding network)
- **CPU usage**: 20-40% (software encoding, 1080p@30fps)
- **Memory usage**: 200-400MB

These targets may vary based on hardware and network conditions.