Update Training API documentation

docs.json

@@ -1132,10 +1132,7 @@
                 "pages": [
                   "training/api-reference"
                 ],
-                "openapi": {
-                  "source": "training/api-reference/openapi.json",
-                  "directory": "training/api-reference"
-                }
+                "openapi": "training/api-reference/openapi.json"
               }
             ]
           },
@@ -2636,4 +2633,4 @@
     }
   ],
   "baseUrl": "https://docs.wandb.ai"
-}
+}

scripts/reference-generation/training/README.md

@@ -95,6 +95,16 @@ After running the scripts:
    grep "training/api-reference" docs.json
    ```
 
+## Known Issues and Workarounds
+
+### Missing Tags on Health Endpoints (RESOLVED)
+
+**Issue**: The upstream Training API's OpenAPI spec was missing `tags` fields on the health endpoints (`/v1/health` and `/v1/system-check`). This caused Mintlify to place these endpoints in an incorrect navigation hierarchy with an extra "API Reference" layer in the breadcrumb.
+
+**Status**: ✅ **FIXED** - The upstream fix was deployed on January 21, 2026 via [PR #213](https://github.com/coreweave/serverless-training/pull/213). The health endpoints now correctly include `"tags": ["health"]` in the OpenAPI spec.
+
+**Temporary Workaround**: The `sync_openapi_spec.py` script includes a `patch_spec()` function that automatically adds the missing tags when syncing. This workaround can be removed once the next sync pulls the fixed spec from production.
+
 ## Troubleshooting
 
 ### 502 Bad Gateway Errors

scripts/reference-generation/training/sync_openapi_spec.py

@@ -44,6 +44,31 @@ def spec_hash(spec: dict) -> str:
     return hashlib.sha256(spec_str.encode()).hexdigest()
 
 
+def patch_spec(spec: dict) -> dict:
+    """
+    Apply patches to the OpenAPI spec to fix known issues.
+
+    This is a workaround until the upstream Training API is fixed.
+    """
+    patched = False
+
+    # Fix missing tags on health endpoints
+    health_endpoints = ["/v1/health", "/v1/system-check"]
+    for path in health_endpoints:
+        if path in spec.get("paths", {}):
+            for method in ["get", "post", "put", "delete", "patch"]:
+                if method in spec["paths"][path]:
+                    if "tags" not in spec["paths"][path][method]:
+                        spec["paths"][path][method]["tags"] = ["health"]
+                        print(f"  ✓ Added missing 'health' tag to {method.upper()} {path}")
+                        patched = True
+
+    if patched:
+        print("  ℹ Applied patches to fix upstream OpenAPI spec issues")
+
+    return spec
+
+
 def compare_specs(local_spec: dict, remote_spec: dict) -> Tuple[bool, list]:
     """
     Compare local and remote specs.
@@ -138,6 +163,9 @@ def main():
             print("  ✗ No local spec and couldn't fetch remote spec")
             return 1
 
+    # Apply patches to fix known issues
+    remote_spec = patch_spec(remote_spec)
+
     # Load local spec
     local_spec = load_local_spec(local_spec_path)
 

training/api-reference.mdx

@@ -29,6 +29,7 @@ https://api.training.wandb.ai/v1
 ### chat-completions
 
 - **[POST /v1/chat/completions](https://docs.wandb.ai/training/api-reference/chat-completions/create-chat-completion)** - Create Chat Completion
+- **[POST /v1/chat/completions/](https://docs.wandb.ai/training/api-reference/chat-completions/create-chat-completion)** - Create Chat Completion
 
 ### models
 

training/api-reference/openapi.json

@@ -563,6 +563,9 @@
     },
     "/v1/health": {
       "get": {
+        "tags": [
+          "health"
+        ],
         "summary": "Health Check",
         "operationId": "health_check_v1_health_get",
         "responses": {
@@ -579,6 +582,9 @@
     },
     "/v1/system-check": {
       "get": {
+        "tags": [
+          "health"
+        ],
         "summary": "System Check",
         "description": "Check health of all system components.\n\nReturns:\n    JSON with status of:\n    - api: Always true (if endpoint is reachable)\n    - database: Whether DB connection works\n    - cpu_queue: Success, duration, and any errors\n    - gpu_queue: Success, duration, and any errors\n\nReturns HTTP 503 if any checks fail.",
         "operationId": "system_check_v1_system_check_get",