Update metadata.md

Author: ayush-portkey

metadata.md

@@ -1,12 +1,12 @@
-# Sending metadata along with your requests
+# Sending Metadata Along with Your Requests
 
-You can send custom metadata along with your API requests which can be later be used for filtering and aggregations.
+You can send custom metadata along with your API requests in Portkey, which can be later used for auditing or filtering logs. Portkey provides four predefined keys: `_environment`, `_user`, `_organisation`, and `_prompt`. These predefined keys are indexed and allow for filtering data in Portkey analytics and logs sections. You can still pass any other metadata key you desire, but these four predefined keys will be indexed and will be available for filtering data in Portkey.
 
 ## Proxy Metadata
 
-To include metadata in the proxy requests, you can add a `x-portkey-meta` header with a JSON string in it. Portkey will parse the JSON object and make it available for filtering.
+To include metadata in the proxy requests, you can add an `x-portkey-meta` header with a JSON string containing your metadata. Portkey will parse the JSON object and make it available for filtering.
 
-For the Javascript library, here's the sample code
+For the JavaScript library, use the following sample code:
 
 ```javascript
 import { Configuration, OpenAIApi } from "openai";
@@ -16,54 +16,57 @@ const configuration = new Configuration({
     basePath: "https://api.portkey.ai/v1/proxy", // Replace openai with portkey's endpoint
     baseOptions: {
       headers: {
-        "x-portkey-api-key": "<YOUR PORTKEY API KEY>", // Can be got from your account
-        "x-portkey-mode": "proxy openai", // Tells portkey to proxy your request to openai
-        "x-portkey-meta": JSON.stringify({"foo": "abc", "bar": "def"}) // Enables filtering on `foo` and `bar`
+        "x-portkey-api-key": "<YOUR PORTKEY API KEY>", // Can be obtained from your account
+        "x-portkey-mode": "proxy openai", // Instructs Portkey to proxy your request to OpenAI
+        "x-portkey-meta": JSON.stringify({"_environment": "production", "foo": "abc", "bar": "def"}) //Enables filtering on _environment
       }
     }
 });
 const openai = new OpenAIApi(configuration);
 ```
 
-For Python
+For Python:
+
 ```python
 import json
 
-openai.api_base = "https://api.portkey.ai/v1/proxy" # Replace openai with portkey's endpoint
+openai.api_base = "https://api.portkey.ai/v1/proxy"  # Replace openai with Portkey's endpoint
 
 response = openai.ChatCompletion.create(
-  model="gpt-3.5-turbo",
-  messages=[
+    model="gpt-3.5-turbo",
+    messages=[
         {"role": "system", "content": "You are a helpful assistant"},
-        {"role": "user", "content": "Create a 5 day trip plan for Laos."},
+        {"role": "user", "content": "Create a 5-day trip plan for Laos."},
     ],
-  temperature=0.2,
-  headers={ # Add portkey headers for auth and proxy mode
-    "x-portkey-api-key": "<YOUR PORTKEY API KEY>",
-    "x-portkey-mode": "proxy openai",
-    "x-portkey-meta": json.dumps({"foo": "abc", "bar": "def"}) # Enables filtering on `foo` and `bar`
-  }
+    temperature=0.2,
+    headers={  # Add Portkey headers for authentication and proxy mode
+        "x-portkey-api-key": "<YOUR PORTKEY API KEY>",
+        "x-portkey-mode": "proxy openai",
+        "x-portkey-meta": json.dumps({"_environment": "production", "foo": "abc", "bar": "def"})  # Enables filtering on _environment
+    }
 )
 ```
 
-### 2.3 Using Langchain
-Add the following code in your langchain initialisation to enable portkey.
+### Using Langchain
+
+To enable Portkey in your Langchain initialization, add the following code:
 
 ```python
 from langchain.llms import OpenAI
 import openai
 import json
+
 openai.api_base = "https://api.portkey.ai/v1/proxy"
 
 llm = OpenAI(temperature=0.2, headers={
     "x-portkey-api-key": "<YOUR PORTKEY API KEY>",
     "x-portkey-mode": "proxy openai",
-    "x-portkey-meta": json.dumps({"foo": "abc", "bar": "def"}) # Enables filtering on `foo` and `bar`
+    "x-portkey-meta": json.dumps({"_environment": "production", "foo": "abc", "bar": "def"})  # Enables filtering on _environment
 })
-text = "Create a 5 day trip plan for Laos."
+text = "Create a 5-day trip plan for Laos."
 print(llm(text))
 ```
 
-## Model completion metadata
+#### NOTE : When using the **_user** predefined key, the following behavior applies:
 
-_Coming soon_
+If you pass the `user` key in the OpenAI request body, it will be automatically stored in `_user`. If both the OpenAI request body `user` key and the metadata `_user` key are passed, the metadata `_user` key will be stored.