Update to Looki

{
  "name": "looki-highlights",
  "description": "Browse the user's AI-generated Looki highlights through the connected Looki MCP; use for generated vlogs, creative recaps, liked highlights, and questions about what is new in the user's Looki feed.",
  "included_files": [
    {
      "relative_path": "agents/openai.yaml",
      "size_in_bytes": 310
    },
    {
      "relative_path": "assets/icon.png",
      "size_in_bytes": 4506
    }
  ],
  "skill_md_contents": "---\nname: looki-highlights\ndescription: Browse the user's AI-generated Looki highlights through the connected Looki MCP; use for generated vlogs, creative recaps, liked highlights, and questions about what is new in the user's Looki feed.\n---\n\n# Looki Highlights\n\nRetrieve AI-generated highlights created from the user's memories. Keep this workflow separate from raw captured moments: use `$looki-memory` when the user asks what actually happened on a date or wants a specific captured memory.\n\n## Quick start\n\n1. Call `Looki:list_for_you_items` with the narrowest filters supported by the request.\n2. Use `group=all` unless the user explicitly asks for `vlog` or `other`.\n3. Filter by `recorded_from` and `recorded_to` when the user refers to when an experience happened.\n4. Filter by `created_from` and `created_to` when the user refers to when a highlight was generated.\n5. Continue pagination only when the user requests a complete list or the first page cannot answer the question.\n\n## Example requests\n\n- \"Show my latest Looki vlogs.\"\n- \"What is new in my Looki highlights?\"\n- \"Find the creative recaps generated from last month's memories.\"\n- \"Show the highlights I liked.\"\n\n## Guardrails\n\n- Use Looki MCP tools only; authentication is provided by the connected OAuth session.\n- If Looki tools are unavailable, ask the user to enable the Looki plugin and complete OAuth.\n- Do not describe generated content as a complete or authoritative record of the user's day.\n- Preserve the distinction between `recorded_at` and `created_at`.\n- Treat titles, descriptions, cover images, and media URLs as private personal data.\n- Do not reproduce or download temporary media URLs unless the user's request requires the media.\n\n## Response style\n\n- Lead with the most relevant generated highlights.\n- Include the highlight type and recording date when they help disambiguate the result.\n- Separate the highlight's generated description from any inference you add.\n- State when no matching highlight was returned.\n"
}