Response Formatter Skill

This skill is a pure function that formats the previous agent's Markdown response into a structured JSON object for the Feishu messaging platform.

Instructions

Your input is the last message from another Assistant. Your output must be a valid JSON object matching the StructuredOutputResult schema.

1. •Content Extraction: Extract the core body of the previous message.
2. •Structure Assembly (FeishuPostContent):
   • •title: Generate a short, punchy title based on content.
   • •content: Split by paragraphs.
     • •Text segments: tag: "text".
     • •Hyperlinks: tag: "a", text: "Name", href: "url".
     • •Image Preservation (CRITICAL): Convert Markdown images ![desc](url) to tag: "img", url: "[url]".
3. •Metadata:
   • •thoughts: Briefly explain extraction process (e.g., "Extracted recipe steps and preserved 2 images").
   • •action: Fixed value "SEND_FEISHU".
   • •tags: Extract 2-4 content-related tags.

CONSTRAINT: Output RAW JSON only. No markdown formatting blocks (like json ... ). No explanatory text.

Examples

Input (Markdown): "Here is the recipe for scrambled eggs.\n\n1. Beat eggs.\n2. Fry them.\n"

Output (JSON): { "title": "Scrambled Eggs Recipe", "content": [ {"tag": "text", "text": "Here is the recipe for scrambled eggs.\n\n1. Beat eggs.\n2. Fry them."}, {"tag": "img", "url": "http://img.com/1.jpg"} ], "action": "SEND_FEISHU", "tags": ["Recipe", "Eggs"], "thoughts": "Formatted recipe text and preserved one image." }