legion-llm 0.3.4 → 0.3.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: bd0b530095616abc383dcd06473a6c435753f021458c76c981da1d4e98583a5f
-  data.tar.gz: 21d8645355c14d591891c3484ca90957e99b0cb376b115eb61dd61f3e0721800
+  metadata.gz: e3e10dfcd60fe722290bec30017671fb261f3baf151b416c82defb082d4445f4
+  data.tar.gz: 0c0b062649f8ede281fada374681d551d437126a2efa0e604a069610e14b7069
 SHA512:
-  metadata.gz: 53f3b6bd09f86625986e6f9d5c53f665e000e71d78dc5db36d599f1b5e5d7267d40ca1a2fe1e9f2b48cc54fb7ab6d272108869a02b327ff9669f978b83280e71
-  data.tar.gz: 381d707e3bdb75a1cf87d82404dc842140f98fa4bb5091e5a837685235327684b800de977efcd24857bb0ab8ab5bc75d42fdc22364aa4b024d6adf8f27cdab65
+  metadata.gz: eeb2cd074c2eb1c3b63ccb7644adbcf7cac6bab62f8d5cc966e318b2185267ab73fae920726b83e1f72cbf8753ba1245ab84ae914baa092aebdbef08e0548cd3
+  data.tar.gz: ccc52360f869421100f0bbda503570168f2f7eb86c5fda9069e6ee29bdaa4c60553d202a1ad2e2109e98ba451b9abf9c00dc8210b50cf5d0a925192cada2ab9d

data/CHANGELOG.md

@@ -1,5 +1,12 @@
 # Legion LLM Changelog
 
+## [0.3.5] - 2026-03-18
+
+### Added
+- Gateway integration: `chat`, `embed`, `structured` delegate to `lex-llm-gateway` when loaded for automatic metering and fleet dispatch
+- `chat_direct`, `embed_direct`, `structured_direct` methods bypass gateway (used by gateway runners to avoid recursion)
+- Gateway integration spec (8 examples)
+
 ## [0.3.4] - 2026-03-18
 
 ### Added

data/Gemfile

@@ -4,6 +4,8 @@ source 'https://rubygems.org'
 
 gemspec
 
+gem 'lex-llm-gateway', path: '../extensions-core/lex-llm-gateway' if File.directory?('../extensions-core/lex-llm-gateway')
+
 group :test do
   gem 'rake'
   gem 'rspec'

data/lib/legion/llm/version.rb

@@ -2,6 +2,6 @@
 
 module Legion
   module LLM
-    VERSION = '0.3.4'
+    VERSION = '0.3.5'
   end
 end

data/lib/legion/llm.rb

@@ -9,6 +9,12 @@ require 'legion/llm/compressor'
 require 'legion/llm/quality_checker'
 require 'legion/llm/escalation_history'
 
+begin
+  require 'legion/extensions/llm/gateway'
+rescue LoadError
+  nil
+end
+
 module Legion
   module LLM
     class EscalationExhausted < StandardError; end
@@ -50,20 +56,24 @@ module Legion
         end
       end
 
-      # Create a new chat session
-      # @param model [String] model ID (e.g., "us.anthropic.claude-sonnet-4-6-v1")
-      # @param provider [Symbol] provider slug (e.g., :bedrock, :anthropic)
-      # @param intent [Hash, nil] routing intent (capability, privacy, etc.)
-      # @param tier [Symbol, nil] explicit tier override — skips rule matching
-      # @param escalate [Boolean, nil] enable escalation retry loop (nil = auto from settings)
-      # @param max_escalations [Integer, nil] max escalation attempts override
-      # @param quality_check [Proc, nil] custom quality check callable
-      # @param message [String, nil] message to send (required for escalation)
-      # @param kwargs [Hash] additional options passed to RubyLLM.chat
-      # @return [RubyLLM::Chat]
-      # TODO: fleet tier dispatch via Transport (Phase 3)
+      # Create a new chat session — delegates to lex-llm-gateway when available
+      # for automatic metering and fleet dispatch
       def chat(model: nil, provider: nil, intent: nil, tier: nil, escalate: nil,
                max_escalations: nil, quality_check: nil, message: nil, **)
+        if gateway_loaded? && message
+          return gateway_chat(model: model, provider: provider, intent: intent,
+                              tier: tier, message: message, escalate: escalate,
+                              max_escalations: max_escalations, quality_check: quality_check, **)
+        end
+
+        chat_direct(model: model, provider: provider, intent: intent, tier: tier,
+                    escalate: escalate, max_escalations: max_escalations,
+                    quality_check: quality_check, message: message, **)
+      end
+
+      # Direct chat bypassing gateway — used by gateway runners to avoid recursion
+      def chat_direct(model: nil, provider: nil, intent: nil, tier: nil, escalate: nil,
+                      max_escalations: nil, quality_check: nil, message: nil, **)
         escalate = escalation_enabled? if escalate.nil?
 
         if escalate && message
@@ -77,11 +87,15 @@ module Legion
         end
       end
 
-      # Generate embeddings via Embeddings module
-      # @param text [String, Array<String>] text to embed
-      # @param model [String] embedding model ID
-      # @return [Hash] { vector:, model:, dimensions:, tokens: }
+      # Generate embeddings — delegates to gateway when available
       def embed(text, **)
+        return Legion::Extensions::LLM::Gateway::Runners::Inference.embed(text: text, **) if gateway_loaded?
+
+        embed_direct(text, **)
+      end
+
+      # Direct embed bypassing gateway
+      def embed_direct(text, **)
         require 'legion/llm/embeddings'
         Embeddings.generate(text: text, **)
       end
@@ -94,11 +108,19 @@ module Legion
         Embeddings.generate_batch(texts: texts, **)
       end
 
-      # Generate structured JSON output from LLM
-      # @param messages [Array<Hash>] conversation messages
-      # @param schema [Hash] JSON schema to enforce
-      # @return [Hash] { data:, raw:, model:, valid: }
+      # Generate structured JSON output — delegates to gateway when available
       def structured(messages:, schema:, **)
+        if gateway_loaded?
+          return Legion::Extensions::LLM::Gateway::Runners::Inference.structured(
+            messages: messages, schema: schema, **
+          )
+        end
+
+        structured_direct(messages: messages, schema: schema, **)
+      end
+
+      # Direct structured bypassing gateway
+      def structured_direct(messages:, schema:, **)
         require 'legion/llm/structured_output'
         StructuredOutput.generate(messages: messages, schema: schema, **)
       end
@@ -113,6 +135,14 @@ module Legion
 
       private
 
+      def gateway_loaded?
+        defined?(Legion::Extensions::LLM::Gateway::Runners::Inference)
+      end
+
+      def gateway_chat(**)
+        Legion::Extensions::LLM::Gateway::Runners::Inference.chat(**)
+      end
+
       def chat_single(model:, provider:, intent:, tier:, **kwargs)
         if (intent || tier) && Router.routing_enabled?
           resolution = Router.resolve(intent: intent, tier: tier, model: model, provider: provider)

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: legion-llm
 version: !ruby/object:Gem::Version
-  version: 0.3.4
+  version: 0.3.5
 platform: ruby
 authors:
 - Esity