From 617dc0ae33b28db1743143d660f03fad9bd07eee Mon Sep 17 00:00:00 2001
From: Derek Meegan <derek.j.meegan@gmail.com>
Date: Sat, 11 Oct 2025 17:29:47 -0700
Subject: [PATCH 1/3] fix raw json schema support in extract

---
 stagehand/handlers/extract_handler.py | 4 +---
 stagehand/llm/inference.py            | 7 ++++---
 2 files changed, 5 insertions(+), 6 deletions(-)

diff --git a/stagehand/handlers/extract_handler.py b/stagehand/handlers/extract_handler.py
index 8af621c..5fa2421 100644
--- a/stagehand/handlers/extract_handler.py
+++ b/stagehand/handlers/extract_handler.py
@@ -151,9 +151,7 @@ async def extract(
 
         processed_data_payload = raw_data_dict  # Default to the raw dictionary
 
-        if schema and isinstance(
-            raw_data_dict, dict
-        ):  # schema is the Pydantic model type
+        if schema and isinstance(schema, type) and issubclass(schema, BaseModel):
             # Try direct validation first
             try:
                 validated_model_instance = schema.model_validate(raw_data_dict)
diff --git a/stagehand/llm/inference.py b/stagehand/llm/inference.py
index b438883..64d8597 100644
--- a/stagehand/llm/inference.py
+++ b/stagehand/llm/inference.py
@@ -169,11 +169,12 @@ async def extract(
     start_time = time.time()
 
     # Determine if we need to use schema-based response format
-    # TODO: if schema is json, return json
     response_format = {"type": "json_object"}
     if schema:
-        # If schema is a Pydantic model, use it directly
-        response_format = schema
+        if isinstance(schema, dict):
+            response_format = {"type": "json_schema", "json_schema": {"name": "extraction_schema", "strict": False, "schema": schema}}
+        else:
+            response_format = schema
 
     # Call the LLM with appropriate parameters
     try:

From c3b523711ecd8f779aedc058e6432d489b4e87ae Mon Sep 17 00:00:00 2001
From: Derek Meegan <derek@browserbase.com>
Date: Tue, 14 Oct 2025 08:47:11 -0700
Subject: [PATCH 2/3] formatting

---
 stagehand/llm/inference.py | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/stagehand/llm/inference.py b/stagehand/llm/inference.py
index 64d8597..5c9d122 100644
--- a/stagehand/llm/inference.py
+++ b/stagehand/llm/inference.py
@@ -172,7 +172,14 @@ async def extract(
     response_format = {"type": "json_object"}
     if schema:
         if isinstance(schema, dict):
-            response_format = {"type": "json_schema", "json_schema": {"name": "extraction_schema", "strict": False, "schema": schema}}
+            response_format = {
+                "type": "json_schema",
+                "json_schema": {
+                    "name": "extraction_schema",
+                    "strict": False,
+                    "schema": schema,
+                },
+            }
         else:
             response_format = schema
 

From 73e828ca07f214194de2271e5dbd49275f84bfd4 Mon Sep 17 00:00:00 2001
From: Derek Meegan <derek@browserbase.com>
Date: Tue, 14 Oct 2025 11:51:18 -0700
Subject: [PATCH 3/3] add changeset

---
 .changeset/fortunate-friendly-avocet.md | 5 +++++
 1 file changed, 5 insertions(+)
 create mode 100644 .changeset/fortunate-friendly-avocet.md

diff --git a/.changeset/fortunate-friendly-avocet.md b/.changeset/fortunate-friendly-avocet.md
new file mode 100644
index 0000000..5f1ee44
--- /dev/null
+++ b/.changeset/fortunate-friendly-avocet.md
@@ -0,0 +1,5 @@
+---
+"stagehand": patch
+---
+
+Fix ability to pass raw JSON to Extract schema