Ensuring openrouter headers are being set properly

lib/active_agent/generation_provider/anthropic_provider.rb

@@ -36,9 +36,12 @@ def generate(prompt)
       end
 
       def chat_prompt(parameters: prompt_parameters)
-        parameters[:stream] = provider_stream if prompt.options[:stream] || config["stream"]
+        if prompt.options[:stream] || config["stream"]
+          parameters[:stream] = provider_stream
+          @streaming_request_params = parameters
+        end
 
-        chat_response(@client.messages(parameters: parameters))
+        chat_response(@client.messages(parameters: parameters), parameters)
       end
 
       protected
@@ -120,7 +123,7 @@ def convert_role(role)
         end
       end
 
-      def chat_response(response)
+      def chat_response(response, request_params = nil)
         return @response if prompt.options[:stream]
 
         content = response["content"].first["text"]
@@ -137,7 +140,8 @@ def chat_response(response)
         @response = ActiveAgent::GenerationProvider::Response.new(
           prompt: prompt,
           message: message,
-          raw_response: response
+          raw_response: response,
+          raw_request: request_params
         )
       end
 

lib/active_agent/generation_provider/open_ai_provider.rb

@@ -69,7 +69,12 @@ def process_stream_chunk(chunk, message, agent_stream)
         elsif chunk.dig("choices", 0, "delta", "tool_calls") && chunk.dig("choices", 0, "delta", "role")
           message = handle_message(chunk.dig("choices", 0, "delta"))
           prompt.messages << message
-          @response = ActiveAgent::GenerationProvider::Response.new(prompt:, message:)
+          @response = ActiveAgent::GenerationProvider::Response.new(
+            prompt:,
+            message:,
+            raw_response: chunk,
+            raw_request: @streaming_request_params
+          )
         end
 
         if chunk.dig("choices", 0, "finish_reason")
@@ -92,18 +97,23 @@ def format_image_content(message)
       # The format_tools method comes from ToolManagement module
       # The provider_messages method comes from MessageFormatting module
 
-      def chat_response(response)
+      def chat_response(response, request_params = nil)
         return @response if prompt.options[:stream]
         message_json = response.dig("choices", 0, "message")
         message_json["id"] = response.dig("id") if message_json["id"].blank?
         message = handle_message(message_json)
 
         update_context(prompt: prompt, message: message, response: response)
 
-        @response = ActiveAgent::GenerationProvider::Response.new(prompt: prompt, message: message, raw_response: response)
+        @response = ActiveAgent::GenerationProvider::Response.new(
+          prompt: prompt,
+          message: message,
+          raw_response: response,
+          raw_request: request_params
+        )
       end
 
-      def responses_response(response)
+      def responses_response(response, request_params = nil)
         message_json = response["output"].find { |output_item| output_item["type"] == "message" }
         message_json["id"] = response.dig("id") if message_json["id"].blank?
 
@@ -116,7 +126,12 @@ def responses_response(response)
           content_type: prompt.output_schema.present? ? "application/json" : "text/plain",
         )
 
-        @response = ActiveAgent::GenerationProvider::Response.new(prompt: prompt, message: message, raw_response: response)
+        @response = ActiveAgent::GenerationProvider::Response.new(
+          prompt: prompt,
+          message: message,
+          raw_response: response,
+          raw_request: request_params
+        )
       end
 
       def handle_message(message_json)
@@ -133,13 +148,16 @@ def handle_message(message_json)
       # handle_actions is now provided by ToolManagement module
 
       def chat_prompt(parameters: prompt_parameters)
-        parameters[:stream] = provider_stream if prompt.options[:stream] || config["stream"]
-        chat_response(@client.chat(parameters: parameters))
+        if prompt.options[:stream] || config["stream"]
+          parameters[:stream] = provider_stream
+          @streaming_request_params = parameters
+        end
+        chat_response(@client.chat(parameters: parameters), parameters)
       end
 
       def responses_prompt(parameters: responses_parameters)
         # parameters[:stream] = provider_stream if prompt.options[:stream] || config["stream"]
-        responses_response(@client.responses.create(parameters: parameters))
+        responses_response(@client.responses.create(parameters: parameters), parameters)
       end
 
       def responses_parameters(model: @prompt.options[:model] || @model_name, messages: @prompt.messages, temperature: @prompt.options[:temperature] || @config["temperature"] || 0.7, tools: @prompt.actions, structured_output: @prompt.output_schema)
@@ -158,14 +176,20 @@ def embeddings_parameters(input: prompt.message.content, model: "text-embedding-
         }
       end
 
-      def embeddings_response(response)
+      def embeddings_response(response, request_params = nil)
         message = ActiveAgent::ActionPrompt::Message.new(content: response.dig("data", 0, "embedding"), role: "assistant")
 
-        @response = ActiveAgent::GenerationProvider::Response.new(prompt: prompt, message: message, raw_response: response)
+        @response = ActiveAgent::GenerationProvider::Response.new(
+          prompt: prompt,
+          message: message,
+          raw_response: response,
+          raw_request: request_params
+        )
       end
 
       def embeddings_prompt(parameters:)
-        embeddings_response(@client.embeddings(parameters: embeddings_parameters))
+        params = embeddings_parameters
+        embeddings_response(@client.embeddings(parameters: params), params)
       end
     end
   end

lib/active_agent/generation_provider/open_router_provider.rb

@@ -38,7 +38,7 @@ def initialize(config)
           uri_base: "https://openrouter.ai/api/v1",
           access_token: @access_token,
           log_errors: Rails.env.development?,
-          default_headers: openrouter_headers
+          extra_headers: openrouter_headers
         )
       end
 

lib/active_agent/generation_provider/response.rb

@@ -3,28 +3,24 @@
 module ActiveAgent
   module GenerationProvider
     class Response
-      attr_reader :message, :prompt, :raw_response
+      attr_reader :message, :prompt, :raw_response, :raw_request
       attr_accessor :metadata
 
-      def initialize(prompt:, message: nil, raw_response: nil, metadata: nil)
+      def initialize(prompt:, message: nil, raw_response: nil, raw_request: nil, metadata: nil)
         @prompt = prompt
         @message = message || prompt.message
         @raw_response = raw_response
+        @raw_request = sanitize_request(raw_request)
         @metadata = metadata || {}
       end
 
       # Extract usage statistics from the raw response
       def usage
         return nil unless @raw_response
 
-        # OpenAI/OpenRouter format
+        # Most providers store usage in the same format
         if @raw_response.is_a?(Hash) && @raw_response["usage"]
           @raw_response["usage"]
-        # Anthropic format
-        elsif @raw_response.is_a?(Hash) && @raw_response["usage"]
-          @raw_response["usage"]
-        else
-          nil
         end
       end
 
@@ -40,6 +36,40 @@ def completion_tokens
       def total_tokens
         usage&.dig("total_tokens")
       end
+
+      private
+
+      def sanitize_request(request)
+        return nil if request.nil?
+        return request unless request.is_a?(Hash)
+
+        # Deep clone the request to avoid modifying the original
+        sanitized = request.deep_dup
+
+        # Sanitize any string values in the request
+        sanitize_hash_values(sanitized)
+      end
+
+      def sanitize_hash_values(hash)
+        hash.each do |key, value|
+          case value
+          when String
+            # Use ActiveAgent's sanitize_credentials to replace sensitive data
+            hash[key] = ActiveAgent.sanitize_credentials(value)
+          when Hash
+            sanitize_hash_values(value)
+          when Array
+            value.each_with_index do |item, index|
+              if item.is_a?(String)
+                value[index] = ActiveAgent.sanitize_credentials(item)
+              elsif item.is_a?(Hash)
+                sanitize_hash_values(item)
+              end
+            end
+          end
+        end
+        hash
+      end
     end
   end
 end

test/dummy/config/credentials.yml.enc

@@ -1 +1 @@
-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--85+CXyjvfbFJi6BM--CHQakeLJ4Z0qHE/8QWnxVQ==
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--AgUCVbFPGIA1BP05--YI0wFyd8L6DcXLi7cDU6Cg==

test/generation_provider/anthropic_raw_request_test.rb

@@ -0,0 +1,155 @@
+require "test_helper"
+require "active_agent/generation_provider/anthropic_provider"
+require "active_agent/action_prompt/prompt"
+
+module ActiveAgent
+  module GenerationProvider
+    class AnthropicRawRequestTest < ActiveSupport::TestCase
+      setup do
+        @config = {
+          "api_key" => "test-key",
+          "model" => "claude-3-opus-20240229"
+        }
+        @provider = AnthropicProvider.new(@config)
+
+        @prompt = ActiveAgent::ActionPrompt::Prompt.new(
+          messages: [
+            ActiveAgent::ActionPrompt::Message.new(
+              content: "Hello, Claude!",
+              role: "user"
+            )
+          ],
+          actions: [],
+          options: {},
+          output_schema: nil
+        )
+      end
+
+      test "chat_response includes raw_request when provided" do
+        mock_response = {
+          "id" => "msg-123",
+          "content" => [
+            {
+              "type" => "text",
+              "text" => "Hello! I'm Claude. How can I assist you today?"
+            }
+          ],
+          "stop_reason" => "end_turn",
+          "usage" => {
+            "input_tokens" => 10,
+            "output_tokens" => 12
+          }
+        }
+
+        request_params = {
+          model: "claude-3-opus-20240229",
+          messages: [ { role: "user", content: "Hello, Claude!" } ],
+          max_tokens: 1024,
+          temperature: 0.7
+        }
+
+        @provider.instance_variable_set(:@prompt, @prompt)
+        response = @provider.send(:chat_response, mock_response, request_params)
+
+        assert_not_nil response
+        assert_equal request_params, response.raw_request
+        assert_equal mock_response, response.raw_response
+      end
+
+      test "chat_response with tool use includes raw_request" do
+        mock_response = {
+          "id" => "msg-456",
+          "content" => [
+            {
+              "type" => "text",
+              "text" => "I'll help you with that calculation."
+            },
+            {
+              "type" => "tool_use",
+              "id" => "tool-789",
+              "name" => "calculator",
+              "input" => { "expression" => "2 + 2" }
+            }
+          ],
+          "stop_reason" => "tool_use"
+        }
+
+        request_params = {
+          model: "claude-3-opus-20240229",
+          messages: [ { role: "user", content: "What is 2 + 2?" } ],
+          tools: [
+            {
+              name: "calculator",
+              description: "Performs calculations",
+              input_schema: {
+                type: "object",
+                properties: {
+                  expression: { type: "string" }
+                }
+              }
+            }
+          ],
+          max_tokens: 1024
+        }
+
+        @provider.instance_variable_set(:@prompt, @prompt)
+        response = @provider.send(:chat_response, mock_response, request_params)
+
+        assert_not_nil response
+        assert_equal request_params, response.raw_request
+        assert_equal mock_response, response.raw_response
+        assert response.message.action_requested
+      end
+
+      test "streaming request params are captured" do
+        request_params = {
+          model: "claude-3-opus-20240229",
+          messages: [ { role: "user", content: "Stream test" } ],
+          stream: true,
+          max_tokens: 1024
+        }
+
+        @provider.instance_variable_set(:@prompt, @prompt)
+
+        # Simulate setting streaming params like in chat_prompt
+        @provider.instance_variable_set(:@streaming_request_params, request_params)
+
+        assert_equal request_params, @provider.instance_variable_get(:@streaming_request_params)
+      end
+
+      test "response includes metadata alongside raw_request and raw_response" do
+        mock_response = {
+          "id" => "msg-meta-123",
+          "content" => [
+            {
+              "type" => "text",
+              "text" => "Response with metadata"
+            }
+          ],
+          "stop_reason" => "end_turn",
+          "model" => "claude-3-opus-20240229",
+          "usage" => {
+            "input_tokens" => 5,
+            "output_tokens" => 4
+          }
+        }
+
+        request_params = {
+          model: "claude-3-opus-20240229",
+          messages: [ { role: "user", content: "Test" } ],
+          max_tokens: 100
+        }
+
+        @provider.instance_variable_set(:@prompt, @prompt)
+        response = @provider.send(:chat_response, mock_response, request_params)
+
+        assert_not_nil response
+        assert_equal request_params, response.raw_request
+        assert_equal mock_response, response.raw_response
+
+        # Response should also have metadata
+        assert_instance_of Hash, response.metadata
+      end
+    end
+  end
+end