refactor: split streaming, remove some features to decrease size

Author: mishamyrt

CONTRIBUTING.md

@@ -25,6 +25,11 @@ cd perplexity-web-api-mcp
 cargo build --workspace --all-targets
 ```
 
+To build the optimized MCP binary directly:
+
+- `cargo build --profile dist --bin perplexity-web-api-mcp` builds the smaller `stdio`-only binary.
+- `cargo build --profile dist --bin perplexity-web-api-mcp --features streamable-http` enables the optional Streamable HTTP transport.
+
 #### Windows
 
 Building on Windows requires additional tools because this project depends on BoringSSL (via the `rquest` crate).

Cargo.lock

@@ -15,14 +15,15 @@ license = "MIT"
 [workspace.dependencies]
 perplexity-web-api = { path = "crates/perplexity-web-api" }
 
-tokio = { version = "1.51.0", features = ["rt-multi-thread", "macros", "net"] }
-rquest = { version = "5.1.0", features = [
+tokio = { version = "1.51.0", features = ["rt", "macros", "net"] }
+rquest = { version = "5.1.0", default-features = false, features = [
+    "webpki-roots",
     "cookies",
     "json",
     "stream",
     "multipart",
 ] }
-rquest-util = "2.2.1"
+rquest-util = { version = "2.2.1", default-features = false, features = ["emulation"] }
 serde = { version = "1.0.228", features = ["derive"] }
 serde_json = "1.0.149"
 uuid = { version = "1.23.0", features = ["v4"] }
@@ -34,16 +35,16 @@ regex-lite = "0.1.9"
 base64 = "0.22"
 thiserror = "2.0.18"
 pin-project-lite = "0.2.17"
-rmcp = { version = "1.3.0", features = [
+rmcp = { version = "1.3.0", default-features = false, features = [
     "server",
     "macros",
     "transport-io",
     "schemars",
 ] }
-axum = "0.8.8"
+axum = { version = "0.8.8", default-features = false, features = ["tokio", "http1"] }
 schemars = "1.2.1"
 tracing = "0.1.44"
-tracing-subscriber = { version = "0.3.23", features = ["env-filter"] }
+tracing-subscriber = { version = "0.3.23", default-features = false, features = ["fmt"] }
 
 [workspace.lints.rust]
 unreachable_pub = "warn"

Cargo.toml

@@ -5,7 +5,7 @@ RUN apt-get update && apt-get install -y cmake clang && rm -rf /var/lib/apt/list
 WORKDIR /app
 COPY . .
 
-RUN cargo build --profile dist --bin perplexity-web-api-mcp
+RUN cargo build --profile dist --bin perplexity-web-api-mcp --features streamable-http
 
 FROM gcr.io/distroless/cc-debian12
 

Dockerfile

@@ -153,6 +153,10 @@ I recommend using the one-click install badge at the top of this README for VS C
 codex mcp add perplexity --env PERPLEXITY_SESSION_TOKEN="your-session-token" --env PERPLEXITY_CSRF_TOKEN="your-csrf-token" -- npx -y perplexity-web-api-mcp
 ```
 
+### Building from Source
+
+Source build instructions, including optional cargo features, are documented in [CONTRIBUTING.md](CONTRIBUTING.md).
+
 ### Other MCP Clients
 
 Most clients can be manually configured to use the `mcpServers` wrapper in their configuration file (like Cursor). If your client doesn't work, check its documentation for the correct wrapper format.
@@ -170,6 +174,7 @@ docker run -d \
 ```
 
 The container exposes the MCP server via Streamable HTTP at `http://localhost:8080/mcp`.
+The Docker image is built with `--features streamable-http`; local/source builds need the same feature if you want HTTP transport.
 
 Configure your MCP client to connect:
 
@@ -187,7 +192,7 @@ Configure your MCP client to connect:
 
 | Variable | Default | Description |
 |----------|---------|-------------|
-| `MCP_TRANSPORT` | `streamable-http` | Transport mode. `stdio` or `streamable-http` |
+| `MCP_TRANSPORT` | `streamable-http` | Transport mode. `stdio` or `streamable-http` (requires the `streamable-http` cargo feature) |
 | `MCP_HOST` | `0.0.0.0` | Host address to bind |
 | `MCP_PORT` | `8080` | Port to listen on |
 

README.md

@@ -12,12 +12,16 @@ keywords = { workspace = true }
 name = "perplexity-web-api-mcp"
 path = "src/main.rs"
 
+[features]
+default = []
+streamable-http = ["dep:axum", "rmcp/transport-streamable-http-server"]
+
 [dependencies]
 perplexity-web-api = { workspace = true, features = ["schemars"] }
 
 tokio = { workspace = true, features = ["signal"] }
-rmcp = { workspace = true, features = ["transport-streamable-http-server"] }
-axum = { workspace = true }
+rmcp = { workspace = true }
+axum = { workspace = true, optional = true }
 serde = { workspace = true }
 serde_json = { workspace = true }
 base64 = { workspace = true }

crates/perplexity-web-api-mcp/Cargo.toml

@@ -3,18 +3,17 @@
 mod server;
 
 use perplexity_web_api::{AuthCookies, Client, ReasonModel, SearchModel};
-use rmcp::{
-    ServiceExt,
-    transport::{
-        stdio,
-        streamable_http_server::{StreamableHttpService, session::local::LocalSessionManager},
-    },
-};
+use rmcp::{ServiceExt, transport::stdio};
 use std::{env, env::VarError};
-use tracing_subscriber::{EnvFilter, fmt};
+use tracing_subscriber::fmt;
 
 use crate::server::PerplexityServer;
 
+#[cfg(feature = "streamable-http")]
+use rmcp::transport::streamable_http_server::{
+    StreamableHttpService, session::local::LocalSessionManager,
+};
+
 #[cfg(unix)]
 async fn shutdown_signal() {
     use tokio::signal::unix::{SignalKind, signal};
@@ -80,13 +79,11 @@ where
     }
 }
 
-#[tokio::main]
+#[tokio::main(flavor = "current_thread")]
 async fn main() -> Result<(), Box<dyn std::error::Error>> {
     // Initialize tracing (logs to stderr to not interfere with stdio transport)
     fmt()
-        .with_env_filter(
-            EnvFilter::from_default_env().add_directive(tracing::Level::INFO.into()),
-        )
+        .with_max_level(tracing::Level::INFO)
         .with_writer(std::io::stderr)
         .with_ansi(false)
         .init();
@@ -166,6 +163,7 @@ async fn main() -> Result<(), Box<dyn std::error::Error>> {
                 }
             }
         }
+        #[cfg(feature = "streamable-http")]
         "streamable-http" => {
             let host = optional_env("MCP_HOST")?.unwrap_or_else(|| "0.0.0.0".to_owned());
             let port = optional_env("MCP_PORT")?.unwrap_or_else(|| "8080".to_owned());
@@ -182,9 +180,20 @@ async fn main() -> Result<(), Box<dyn std::error::Error>> {
             tracing::info!("MCP server listening on http://{addr}/mcp");
             axum::serve(listener, app).with_graceful_shutdown(shutdown_signal()).await?;
         }
+        #[cfg(not(feature = "streamable-http"))]
+        "streamable-http" => {
+            return Err(std::io::Error::other(
+                "MCP_TRANSPORT=streamable-http requires building with the `streamable-http` cargo feature",
+            )
+            .into());
+        }
         other => {
+            #[cfg(feature = "streamable-http")]
+            let valid_values = "'stdio', 'streamable-http'";
+            #[cfg(not(feature = "streamable-http"))]
+            let valid_values = "'stdio'";
             return Err(std::io::Error::other(format!(
-                "Unknown MCP_TRANSPORT value: '{other}'. Valid values: 'stdio', 'streamable-http'"
+                "Unknown MCP_TRANSPORT value: '{other}'. Valid values: {valid_values}"
             ))
             .into());
         }

crates/perplexity-web-api-mcp/src/main.rs

@@ -4,7 +4,7 @@
 
 use perplexity_web_api::{Client, SearchRequest, Source};
 
-#[tokio::main]
+#[tokio::main(flavor = "current_thread")]
 async fn main() -> perplexity_web_api::Result<()> {
     println!("Creating Perplexity client...");
 

crates/perplexity-web-api/examples/basic.rs

@@ -5,7 +5,7 @@
 use futures_util::StreamExt;
 use perplexity_web_api::{Client, SearchRequest};
 
-#[tokio::main]
+#[tokio::main(flavor = "current_thread")]
 async fn main() -> perplexity_web_api::Result<()> {
     println!("Creating Perplexity client...");
 

crates/perplexity-web-api/examples/streaming.rs

@@ -4,7 +4,7 @@
 
 use perplexity_web_api::{AuthCookies, Client, SearchMode, SearchModel, SearchRequest};
 
-#[tokio::main]
+#[tokio::main(flavor = "current_thread")]
 async fn main() -> perplexity_web_api::Result<()> {
     println!("=== Authenticated Usage Example ===\n");