Update to Reelyze

{
  "description": "Diagnose why a specific short-form video underperformed by analysing the video itself. Use when given an Instagram, TikTok or YouTube Shorts link and asked why it flopped, how to fix the hook, or where viewers dropped off.",
  "included_files": [],
  "name": "analyze-reel",
  "skill_md_contents": "---\nname: analyze-reel\ndescription: Diagnose why a specific short-form video underperformed by analysing the video itself. Use when given an Instagram, TikTok or YouTube Shorts link and asked why it flopped, how to fix the hook, or where viewers dropped off.\n---\n\n# Diagnose a reel from its link\n\nRuns the video through Reelyze: frame-by-frame vision, transcript, scene cuts,\nand a predicted retention curve. Returns where viewers are predicted to leave\nand what to change.\n\n## Tools\n\n| tool | cost | returns |\n|---|---|---|\n| `get_transcript(url)` | free, metered | the spoken words |\n| `analyze_reel(url)` | **paid plan, spends one analysis** | full report |\n| `explain_drop_off(analysis_id)` | paid plan | per-moment retention diagnosis |\n| `check_job(job_id)` | free | poll a running job |\n\n`get_transcript` and `analyze_reel` are **asynchronous**. They return a\n`job_id`; poll `check_job` until `status` is `completed` or `failed`. A full\nanalysis takes minutes, not seconds. Tell the person it is running rather than\npolling in silence.\n\n## Spend the paid call deliberately\n\n`analyze_reel` consumes one analysis from the person's plan allowance, the same\nallowance the Reelyze dashboard spends. Before calling it:\n\n- confirm this is the video they want analysed, if more than one link is in play\n- if they only asked what was said, `get_transcript` is free and sufficient\n- if the call returns 402, their plan has not billed yet. Say so and link\n  https://getreelyze.com/billing rather than retrying.\n\n## Read the report in the order that matters for reach\n\nThis order is not cosmetic. It is the order these levers actually move\ndistribution, and a report read out of order sends people to fix the wrong\nthing.\n\n1. **Skip rate, the hook, first 3 seconds.** The single biggest lever. If the\n   opening loses people, nothing later in the video gets a chance.\n2. **Watch time, retention through the rest.** A *separate* lever, not a\n   restatement of the hook. A video can survive the first 3 seconds and still\n   collapse at 5 to 8 seconds, and low average watch time stops the algorithm\n   expanding reach even though the opening worked.\n3. shares\n4. likes\n5. saves\n6. reposts\n7. comments\n\nLead with whichever of 1 and 2 the report actually implicates. Do not invent a\nmid-video problem for a video that holds its middle.\n\n## One caveat you must not get wrong\n\nThe reported `skip_rate` is **not** a 3-second measurement despite the name.\nInstagram's API exposes no true 3-second skip rate, so it is derived from\naverage watch time across the whole video. That one number therefore spans both\nlevers 1 and 2 and cannot separate \"left at 2 seconds\" from \"left at 15\nseconds\". Only the retention graph from `explain_drop_off` shows the shape of\nthe drop.\n\nSo never assert that a high skip rate proves the opening was weak. It is\nconsistent with a weak opening and equally consistent with a strong opening and\na collapse at 8 seconds.\n\n## Give fixes that can be executed while editing\n\nA diagnosis the person cannot act on is half an answer. For each moment, name\nthe concrete on-screen change: the exact caption words to overlay and where, a\nhard cut or zoom, a progress cue, b-roll, a pattern interrupt. \"Improve the\nhook\" is not a fix; \"put 'I lost $4,000 on this' on screen at 0:00 and cut the\nestablishing shot\" is.\n\n## Links this accepts\n\nInstagram reels, TikTok, YouTube Shorts. Share strings are tolerated: a pasted\nTikTok link with the trailing \"This post is shared via TikTok\" caption is\nnormalised before download, so pass the string as given rather than editing it.\n"
}