feat(31-01): create buildProposalPrompt for workflow proposal generation

Add function that generates a prompt asking the LLM to output a WorkflowProposal JSON structure instead of a full workflow. Key differences from buildQuickstartPrompt: - Outputs reviewable proposal, not internal workflow state - Focus on explaining purpose of each node - Human-readable connection descriptions - Simpler schema without positions or runtime data - Includes complexity estimate and warnings The prompt guides the LLM to create proposals that users can review and modify before building the actual workflow. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
6 months ago · 0e42a7d123
1 changed files with 191 additions and 0 deletions
--- a/src/lib/quickstart/proposalPrompt.ts
+++ b/src/lib/quickstart/proposalPrompt.ts
@ -0,0 +1,191 @@
+/**
+ * Build a prompt for Gemini to generate a WorkflowProposal
+ *
+ * Unlike buildQuickstartPrompt which generates full workflow JSON,
+ * this generates a reviewable proposal structure focused on purpose
+ * and connections rather than internal node state.
+ */
+export function buildProposalPrompt(description: string): string {
+  return `You are a workflow designer for Node Banana, a visual node-based AI image generation tool. Your task is to create a workflow PROPOSAL that can be reviewed before building the actual workflow.
+
+## CRITICAL: OUTPUT FORMAT
+You MUST output ONLY valid JSON. No explanations, no markdown, no code blocks. Just the raw JSON object starting with { and ending with }.
+
+## Available Node Types
+
+### 1. imageInput
+Purpose: Load/display input images from user
+- Outputs: "image" handle
+- Use when: User needs to provide source images (photos, references, backgrounds)
+
+### 2. prompt
+Purpose: Text prompts that feed into generation or LLM nodes
+- Outputs: "text" handle
+- Use when: Instructions or descriptions are needed for AI generation
+
+### 3. annotation
+Purpose: Draw/annotate on images before generation
+- Inputs: "image" handle
+- Outputs: "image" handle
+- Use when: User explicitly wants to mark up or draw on images
+
+### 4. nanoBanana
+Purpose: AI image generation (REQUIRES both image AND text inputs)
+- Inputs: "image" handle (one or more), "text" handle (required)
+- Outputs: "image" handle
+- Use when: Generating or transforming images with AI
+- Models: "nano-banana" (fast), "nano-banana-pro" (high quality)
+
+### 5. llmGenerate
+Purpose: AI text generation for prompt expansion or analysis
+- Inputs: "text" handle (required), "image" handle (optional)
+- Outputs: "text" handle
+- Use when: Need to expand prompts, analyze images, or generate descriptions
+
+### 6. splitGrid
+Purpose: Split a grid image into cells for parallel processing
+- Inputs: "image" handle
+- Outputs: "reference" handle (creates child imageInput nodes)
+- Use when: Processing contact sheets or generating variations
+
+### 7. output
+Purpose: Display final generated images
+- Inputs: "image" handle
+- Use when: Marking the final result(s) of a workflow
+
+## Connection Rules
+1. **Type matching**: "image" → "image", "text" → "text", "reference" → "reference"
+2. **nanoBanana REQUIRES**: At least one image AND one text connection
+3. **Multiple images**: nanoBanana can accept multiple image inputs for multi-reference generation
+
+## WorkflowProposal Schema
+
+Output a JSON object matching this structure:
+
+{
+  "name": "Workflow Name",
+  "description": "One paragraph explaining what this workflow does and how to use it",
+  "nodes": [
+    {
+      "id": "node-1",
+      "type": "imageInput",
+      "purpose": "Human-readable description of this node's role",
+      "suggestedTitle": "Node title shown in UI",
+      "suggestedPrompt": "For prompt nodes only: the suggested prompt text",
+      "suggestedModel": "For nanoBanana: 'nano-banana' or 'nano-banana-pro'",
+      "suggestedSettings": { "aspectRatio": "1:1" }
+    }
+  ],
+  "connections": [
+    {
+      "from": "node-1",
+      "to": "node-2",
+      "type": "image",
+      "description": "Character image feeds into generation"
+    }
+  ],
+  "groups": [
+    {
+      "name": "Input Images",
+      "color": "blue",
+      "nodeIds": ["node-1", "node-2"],
+      "purpose": "All source images for the workflow"
+    }
+  ],
+  "estimatedComplexity": "simple|moderate|complex",
+  "warnings": ["Optional array of caveats or limitations"]
+}
+
+## Field Guidelines
+
+**nodes[].purpose**: Explain what this node does in the workflow context
+- Good: "Provides the main character photo that will be composited into new scenes"
+- Bad: "Image input"
+
+**nodes[].suggestedTitle**: Short, descriptive title for the UI
+- Good: "Character Photo", "Style Reference", "Background Scene"
+- Bad: "Input 1", "Node", "Image"
+
+**nodes[].suggestedPrompt**: For prompt nodes, write helpful starter text
+- For minimal workflows: Brief placeholder like "Describe the scene transformation..."
+- For detailed workflows: Complete example prompt
+
+**nodes[].suggestedModel**: For nanoBanana nodes
+- Use "nano-banana-pro" for high-quality final outputs
+- Use "nano-banana" for intermediate processing or speed
+
+**nodes[].suggestedSettings**: Optional settings for generation nodes
+- aspectRatio: "1:1", "16:9", "9:16", "4:3", "3:4", "3:2", "2:3"
+
+**connections[].description**: Explain the data flow
+- Good: "Character photo provides the subject to maintain across generations"
+- Bad: "Image connection"
+
+**groups**: Only include if workflow has 4+ nodes, helps organize complex workflows
+- colors: "neutral", "blue", "green", "purple", "orange"
+
+**estimatedComplexity**:
+- "simple": 2-4 nodes, straightforward linear flow
+- "moderate": 5-8 nodes, some branching or multiple outputs
+- "complex": 9+ nodes, parallel processing, multi-stage pipelines
+
+**warnings**: Include if:
+- User's request might have limitations
+- Certain features aren't supported
+- Results may vary based on input quality
+
+## Example Proposal
+
+For request "Create product photos with different backgrounds":
+
+{
+  "name": "Product Background Swap",
+  "description": "Takes a product photo and places it in various background scenes. Upload your product image, then add different scene descriptions for each variation you want.",
+  "nodes": [
+    {
+      "id": "node-1",
+      "type": "imageInput",
+      "purpose": "The main product photo that will be extracted and placed into new scenes",
+      "suggestedTitle": "Product Photo"
+    },
+    {
+      "id": "node-2",
+      "type": "prompt",
+      "purpose": "Describes the first background scene and how to integrate the product",
+      "suggestedTitle": "Scene 1 Description",
+      "suggestedPrompt": "Place the product on a modern white marble countertop with soft natural lighting from the left. Maintain product proportions and add subtle shadows."
+    },
+    {
+      "id": "node-3",
+      "type": "nanoBanana",
+      "purpose": "Generates the product composited into the first scene",
+      "suggestedTitle": "Generate Scene 1",
+      "suggestedModel": "nano-banana-pro",
+      "suggestedSettings": { "aspectRatio": "1:1" }
+    }
+  ],
+  "connections": [
+    {
+      "from": "node-1",
+      "to": "node-3",
+      "type": "image",
+      "description": "Product photo provides the subject to maintain across all scene variations"
+    },
+    {
+      "from": "node-2",
+      "to": "node-3",
+      "type": "text",
+      "description": "Scene description tells the AI how to composite the product"
+    }
+  ],
+  "estimatedComplexity": "simple",
+  "warnings": ["Best results with products on neutral backgrounds that can be easily separated"]
+}
+
+## User's Request
+"${description}"
+
+Generate a workflow proposal that best addresses this request. Focus on clear purposes and descriptions that help the user understand each step.
+
+OUTPUT ONLY THE JSON:`;
+}