Update to Dativue

{
  "description": "Use Dativue when a user wants to mine comments or replies from public social posts for audience questions, objections, desires, reactions, and natural language.",
  "included_files": [],
  "name": "dativue-audience-language",
  "skill_md_contents": "---\nname: dativue-audience-language\ndescription: Use Dativue when a user wants to mine comments or replies from public social posts for audience questions, objections, desires, reactions, and natural language.\n---\n\n# Audience-language mining\n\nUse this workflow when the user provides public posts/videos or asks how an audience describes a problem.\n\n1. Gather the exact public post/video URLs. Do not infer or expose private identity attributes.\n2. Retrieve comments directly. Use `research_credit_quote` when the user asks for an estimate or the planned work may exceed 50 credits, and get approval for a higher-cost plan.\n3. Retrieve up to 50 comments/replies per approved URL. Treat the sample as directional, not a representative survey.\n4. Cluster repeated questions, objections, desired outcomes, emotional language, misconceptions, and phrases people naturally use.\n5. Quote only short phrases necessary to demonstrate language patterns and avoid unnecessary usernames or personal information.\n6. Cite the source post for every cluster and recommend message/content angles that are explicitly grounded in the observed comments.\n\nDefault maximum: 24 credits for four source URLs at the highest current comment price.\n"
}