feat: refactoring rendering

frontend/src/components/nodes/agent/descriptor.tsx

@@ -20,5 +20,6 @@ export function getAgentNodeDescriptor(): NodeTypeDescriptor {
     .connectionLabel('prompt/context')
     .withFullscreen()
     .sourceRenderingLogic(agentRenderingLogic)
+    .outputMenuContent(() => null)
     .build()
 }

frontend/src/components/nodes/agent/renderingLogic.ts

@@ -1,10 +1,11 @@
 /**
- * Agent node rendering logic: when the Rendering node runs, it calls the agent API
- * and updates the agent node's output. Update mode is manual (user clicks Run on the renderer).
+ * Agent node rendering logic: implements the "Resolve" step for agent sources.
+ * Calls the agent API and updates the agent node's output. Update mode is manual.
+ * See lib/graph/rendering.ts (IRenderingSource).
  */
 
-import { getConfigContent } from '@/lib/graph/configTypes'
-import type { ResolvedContentResult, SourceRenderingLogicContext } from '@/lib/graph/sourceRenderingLogic'
+import { getConfigContent } from '@/lib/graph/rendering'
+import type { ResolvedContentResult, SourceRenderingLogicContext } from '@/lib/graph/rendering'
 import type { AiConnection } from '@/app/kosmos/KosmosContext'
 
 function serializeNodeForContext(
@@ -69,7 +70,6 @@ function parseReasoningAndOutput(fullMarkdown: string): { reasoning?: string; ou
 
 export const agentRenderingLogic = {
   defaultUpdateMode: 'manual' as const,
-  getOutputMenuDescriptor: (): null => null,
   getResolvedContent: async (context: SourceRenderingLogicContext): Promise<ResolvedContentResult> => {
     const { nodes, sourceNodeId, setNodes, aiConnection, onStreamingStart, onStreamingChunk } = context
     const sourceNode = nodes.find((n) => n.id === sourceNodeId)