fix(proxy): mark converted function tools as non-strict for Responses (#7717)

Anthropic tool schemas treat `required` as an arbitrary subset of
`properties`, but the Responses backend strictifies a function tool when
`strict` is omitted: optional attributes land in `required` and the model
is forced to emit fields the client deliberately left optional (#7713 -
Claude Code Agent always emitting the optional `isolation` and losing
non-worktree teammates).

Set `strict: false` explicitly on the converted function tool in
anthropic_to_responses. The hosted web-search branch is untouched, and
`parameters` still passes through clean_schema unchanged, so the
client-side optionality semantics are preserved end to end. This branch
serves both plain Responses providers and the Codex OAuth backend.

Validation: with the production hunk reverted, the new test asserting
strict=false fails (field absent); restored, the transform_responses
suite and the full --lib run are green.

Co-authored-by: SailingLoong <309827397+SailingLoong@users.noreply.github.com>

src-tauri/src/proxy/providers/transform_responses.rs

@@ -1905,6 +1905,12 @@ pub fn anthropic_to_responses(
                 }
                 let mut response_tool = json!({
                     "type": "function",
+                    // Anthropic 工具 schema 允许可选参数（required 是任意子集），而
+                    // Responses 后端把省略的 `strict` 按 true 处理：可选属性会被
+                    // 严格化进 required，模型被迫输出它本应能省略的字段（#7713 中
+                    // Claude Code Agent 的可选 isolation 被强制输出即此形态）。
+                    // 显式 false 保留客户端 schema 的可选语义。
+                    "strict": false,
                     "name": tool.get("name").and_then(Value::as_str).unwrap_or(""),
                 });
                 // 同 transform.rs：缺失的 description 省略而非输出 null，
@@ -3341,6 +3347,8 @@ mod tests {
 
         let result = anthropic_to_responses(input, None, false, false).unwrap();
         assert_eq!(result["tools"][0]["type"], "function");
+        // 省略 strict 会被 Responses 后端按 true 严格化，可选参数被强制输出（#7713）
+        assert_eq!(result["tools"][0]["strict"], json!(false));
         assert_eq!(result["tools"][0]["name"], "get_weather");
         assert!(result["tools"][0].get("parameters").is_some());
         assert_eq!(result["tools"][0]["parameters"]["type"], json!("object"));
@@ -3352,6 +3360,46 @@ mod tests {
         assert!(result["tools"][0].get("input_schema").is_none());
     }
 
+    #[test]
+    fn test_anthropic_to_responses_marks_function_tools_not_strict() {
+        // #7713：可选参数（required 是任意子集）必须保留可选语义。
+        // 省略 strict 时 Responses 后端按 true 严格化，模型被迫输出 isolation。
+        let input = json!({
+            "model": "gpt-5.6-luna",
+            "max_tokens": 1024,
+            "messages": [{"role": "user", "content": "Call the agent"}],
+            "tools": [{
+                "name": "Agent",
+                "description": "Launch a sub-agent",
+                "input_schema": {
+                    "type": "object",
+                    "properties": {
+                        "description": {"type": "string"},
+                        "prompt": {"type": "string"},
+                        "isolation": {"type": "string", "enum": ["worktree", "remote"]}
+                    },
+                    "required": ["description", "prompt"]
+                }
+            }]
+        });
+
+        let plain = anthropic_to_responses(input.clone(), None, false, false).unwrap();
+        assert_eq!(plain["tools"][0]["strict"], json!(false));
+        // required 保持为原有子集，strict 语义交由后端保留
+        assert_eq!(
+            plain["tools"][0]["parameters"]["required"],
+            json!(["description", "prompt"])
+        );
+
+        // 同一转换分支也服务 Codex OAuth 后端
+        let oauth = anthropic_to_responses(input, None, true, false).unwrap();
+        assert_eq!(oauth["tools"][0]["strict"], json!(false));
+        assert_eq!(
+            oauth["tools"][0]["parameters"]["required"],
+            json!(["description", "prompt"])
+        );
+    }
+
     #[test]
     fn test_anthropic_to_responses_omits_missing_tool_description() {
         let input = json!({