diff --git a/checkpoint.json b/checkpoint.json
new file mode 100644
index 000000000..29a111a1b
--- /dev/null
+++ b/checkpoint.json
@@ -0,0 +1,61 @@
+{
+  "timestamp": "2026-03-19T12:59:56.077751Z",
+  "current_node": "draft",
+  "completed_nodes": [
+    "start",
+    "draft"
+  ],
+  "node_retries": {
+    "start": 1,
+    "draft": 1
+  },
+  "context_values": {
+    "failure_class": "",
+    "internal.fidelity": "compact",
+    "outcome": "success",
+    "graph.rankdir": "LR",
+    "internal.thread_id": "start",
+    "last_stage": "draft",
+    "last_response": "# Greeting\n\nHello! 👋\n\nI'm here to help you test the human gate with auto-approve functionality. This looks like it will involve working with Fabro's workflow orchestration system, likely testing th",
+    "failure_signature": "",
+    "thread.start.current_node": "draft",
+    "response.draft": "# Greeting\n\nHello! 👋\n\nI'm here to help you test the human gate with auto-approve functionality. This looks like it will involve working with Fabro's workflow orchestration system, likely testing the human-in-the-loop interaction features.\n\nI'm ready to assist with:\n- Writing test cases for human gates\n- Configuring auto-approve settings\n- Running workflow tests\n- Debugging any issues that arise\n\nLet me know what specific aspect of the human gate auto-approve feature you'd like to tackle first!",
+    "internal.retry_count.draft": 1,
+    "graph.goal": "Test human gate with auto-approve",
+    "internal.node_visit_count": 1,
+    "current_node": "draft",
+    "internal.run_id": "01KM32WGAZ1XBY4T2YFCKJKYPG",
+    "current.preamble": "Goal: Test human gate with auto-approve\n",
+    "internal.retry_count.start": 1
+  },
+  "logs": [],
+  "node_outcomes": {
+    "start": {
+      "status": "success",
+      "duration_ms": 0
+    },
+    "draft": {
+      "status": "success",
+      "context_updates": {
+        "last_stage": "draft",
+        "last_response": "# Greeting\n\nHello! 👋\n\nI'm here to help you test the human gate with auto-approve functionality. This looks like it will involve working with Fabro's workflow orchestration system, likely testing th",
+        "response.draft": "# Greeting\n\nHello! 👋\n\nI'm here to help you test the human gate with auto-approve functionality. This looks like it will involve working with Fabro's workflow orchestration system, likely testing the human-in-the-loop interaction features.\n\nI'm ready to assist with:\n- Writing test cases for human gates\n- Configuring auto-approve settings\n- Running workflow tests\n- Debugging any issues that arise\n\nLet me know what specific aspect of the human gate auto-approve feature you'd like to tackle first!"
+      },
+      "notes": "Stage completed: draft",
+      "usage": {
+        "model": "claude-haiku-4-5",
+        "input_tokens": 2022,
+        "output_tokens": 119,
+        "cache_read_tokens": 0,
+        "cache_write_tokens": 0,
+        "cost": 0.0020936
+      },
+      "duration_ms": 2081
+    }
+  },
+  "next_node_id": "approve",
+  "node_visits": {
+    "start": 1,
+    "draft": 1
+  }
+}
\ No newline at end of file
diff --git a/nodes/draft/prompt.md b/nodes/draft/prompt.md
new file mode 100644
index 000000000..75d67e5b4
--- /dev/null
+++ b/nodes/draft/prompt.md
@@ -0,0 +1,4 @@
+Goal: Test human gate with auto-approve
+
+
+Write a short greeting message.
\ No newline at end of file
diff --git a/nodes/draft/provider_used.json b/nodes/draft/provider_used.json
new file mode 100644
index 000000000..f79dec8be
--- /dev/null
+++ b/nodes/draft/provider_used.json
@@ -0,0 +1,5 @@
+{
+  "mode": "prompt",
+  "provider": "anthropic",
+  "model": "claude-haiku-4-5"
+}
\ No newline at end of file
diff --git a/nodes/draft/response.md b/nodes/draft/response.md
new file mode 100644
index 000000000..5fcf9a0a3
--- /dev/null
+++ b/nodes/draft/response.md
@@ -0,0 +1,13 @@
+# Greeting
+
+Hello! 👋
+
+I'm here to help you test the human gate with auto-approve functionality. This looks like it will involve working with Fabro's workflow orchestration system, likely testing the human-in-the-loop interaction features.
+
+I'm ready to assist with:
+- Writing test cases for human gates
+- Configuring auto-approve settings
+- Running workflow tests
+- Debugging any issues that arise
+
+Let me know what specific aspect of the human gate auto-approve feature you'd like to tackle first!
\ No newline at end of file
diff --git a/nodes/draft/status.json b/nodes/draft/status.json
new file mode 100644
index 000000000..7a0b387d6
--- /dev/null
+++ b/nodes/draft/status.json
@@ -0,0 +1,6 @@
+{
+  "status": "success",
+  "notes": "Stage completed: draft",
+  "failure_reason": null,
+  "timestamp": "2026-03-19T12:59:56.077512+00:00"
+}
\ No newline at end of file
diff --git a/nodes/start/status.json b/nodes/start/status.json
new file mode 100644
index 000000000..1e08332f0
--- /dev/null
+++ b/nodes/start/status.json
@@ -0,0 +1,6 @@
+{
+  "status": "success",
+  "notes": null,
+  "failure_reason": null,
+  "timestamp": "2026-03-19T12:59:53.994586+00:00"
+}
\ No newline at end of file