diff --git a/agent/Cargo.toml b/agent/Cargo.toml
index 50a2214..e6429bf 100644
--- a/agent/Cargo.toml
+++ b/agent/Cargo.toml
@@ -1,8 +1,12 @@
 [package]
 name = "butterfly-agent"
-version = "0.1.0"
+version = "0.2.0"
 edition = "2021"
-description = "Butterfly Desktop Agent — captures display/audio from a VM and forwards HUD commands (mouse/keyboard)"
+description = "Butterfly Desktop Agent — captures display from a VM, encodes as H.264/JPEG, streams via WebSocket, receives and executes remote control commands"
+
+[features]
+default = ["h264"]
+h264 = ["openh264"]
 
 [dependencies]
 # Async runtime
@@ -12,17 +16,20 @@ tokio = { version = "1", features = ["full"] }
 tokio-tungstenite = { version = "0.24", features = ["connect"] }
 futures-util = "0.3"
 
-# Screen capture (cross-platform: DXGI on Windows, X11 on Linux, CG on macOS)
+# Screen capture (DXGI on Windows, X11 on Linux, CoreGraphics on macOS)
 scrap = "0.5"
 
-# Input simulation (cross-platform mouse/keyboard injection)
+# Input simulation (SendInput on Windows, XTest on Linux, CGEvent on macOS)
 enigo = "0.2"
 
-# Image encoding (BGRA → JPEG)
+# Image encoding (JPEG fallback)
 image = "0.25"
 
-# Base64 encoding for frame data
-base64 = "0.22"
+# H.264 encoding (Cisco OpenH264 — software, cross-platform, ~5-15ms encode)
+openh264 = { version = "0.6", optional = true }
+
+# Conditional compilation
+cfg-if = "1"
 
 # Serialization
 serde = { version = "1", features = ["derive"] }
@@ -44,9 +51,7 @@ anyhow = "1"
 # HTTP client (for REST API session creation)
 reqwest = { version = "0.12", default-features = false, features = ["json", "rustls-tls"] }
 
-# Audio capture (future use)
-# cpal = "0.15"
-
 [profile.release]
 opt-level = 3
 lto = true
+codegen-units = 1