Flesh out readme, Utilize client_id of comfy_runner

README.md

@@ -1,2 +1,11 @@
 # ComfyUI-WorkflowCheckpointing
 Automatically creates checkpoints during workflow execution. If If an workflow is canceled or ComfyUI crashes mid-execution, then these checkpoints are used when the workflow is re-queued to resume execution with minimal progress loss.
+
+## Configuration
+By default, checkpoints are saved localy to a checkpoint folder.
+
+Networked checkpoints backed by the Salad Simple Storage Service can be enabled by setting the `SALAD_ORGANIZATION` environment variable. When deployed to salad, authentication is automatically pulled, but an api key can be applied when testing locally by setting `SALAD_API_KEY`
+
+If a request sent to `/prompt` contains a `client_id`. It is utilized to allow independent caching of executions by multiple users.
+
+To facilitate testing `FORCE_CRASH_AT` can be set to an integer to terminate workflow execution at a given sampling step

workflowcheckpointing.py

@@ -218,11 +218,11 @@ prompt_route = next(filter(lambda x: x.path  == '/prompt' and x.method == 'POST'
 original_post_prompt = prompt_route.handler
 async def post_prompt_remote(request):
     json_data = await request.json()
-    if "prompt" in json_data and "extra_data" in json_data and "SALAD_ORGANIZATION" in os.environ:
-        extra_data = json_data["extra_data"]
+    if "SALAD_ORGANIZATION" in os.environ:
+        extra_data = json_data.get("extra_data", {})
         #NOTE: Rendered obsolete by existing infrastructure, can be pruned
         remote_files = extra_data.get("remote_files", [])
-        uid = extra_data.get("uid", 'local')
+        uid = json_data.get("client_id", 'local')
         checkpoint.uid = uid
         await fetch_remote_files(remote_files, uid=uid)
     return await original_post_prompt(request)