feat(proxy): raise request body limit to 70MB for long LLM contexts

Cargo.toml

@@ -12,7 +12,7 @@ rust-version = "1.86"
 axum = { version = "0.7", features = ["macros"] }
 tokio = { version = "1", features = ["macros", "rt-multi-thread", "time", "signal"] }
 tower = { version = "0.5", features = ["util"] }
-tower-http = { version = "0.5", features = ["cors", "fs", "trace"] }
+tower-http = { version = "0.5", features = ["cors", "fs", "trace", "limit"] }
 hyper = { version = "1.0", features = ["full"] }
 
 # HTTP 客户端（上游转发）

src/main.rs

@@ -210,6 +210,11 @@ async fn main() -> anyhow::Result<()> {
     let state = routes::AppState::new(conn, cfg, crypto);
     let app = routes::router()
         .with_state(state)
+        // 请求体上限：axum 默认 2MB，对超长 LLM 上下文/图片 base64 不够，放宽到 70MB
+        //（rant 2026-08-22T23:20:00；真实限制仍由上游模型 context 窗口裁决）
+        .layer(tower_http::limit::RequestBodyLimitLayer::new(
+            70 * 1024 * 1024,
+        ))
         .layer(tower_http::trace::TraceLayer::new_for_http());
 
     let listener = tokio::net::TcpListener::bind(&addr).await?;