← ReelGenieCONTENT HISTORY

Update to ReelGenie

Snapshot Nov 21, 2026 · 06:12 UTC · version 0.7.1

Collection source: downloaded plugin package.

WHAT CHANGED

First saved snapshot

No earlier snapshot is available to establish a change.

Save ReelGenie and check future changes in Following.

Up to 3 plugins free · no card required.

Compare saved observations

Download comparison JSON
Full technical diff · 0 changed fields
Full snapshot data
{
  "description": "Analyze ReelGenie post reach and engagement, compare recent performers fairly, recover a winner's existing Favourite, and recommend measurable creative experiments.",
  "included_files": [
    {
      "relative_path": "references/comparison-checklist.md",
      "size_in_bytes": 1760
    }
  ],
  "name": "analyze-performance",
  "skill_md_contents": "---\nname: analyze-performance\ndescription: Analyze ReelGenie post reach and engagement, compare recent performers fairly, recover a winner's existing Favourite, and recommend measurable creative experiments.\n---\nUse when the user asks what performed well, requests a period review, or wants to create from a recent winner. Explicit user instructions take precedence. Read-only analysis does not approve posts, alter campaign settings or start an automation.\n\nUse this plugin's bundled connection and resolve requested brands or the default automatically. Read `get_workspace_content_profile`. Use `get_recent_post_performance` for latest per-target counters, `get_latest_post_performance_all_groups` for requested cross-brand comparison, for accessible counters. The Performance panel uses paged `get_performance_review` for saved daily observations, but that tool is app-only: do not assume it can be invoked from chat when absent from the callable catalog. Open the panel through `open_reelgenie` if the user wants its chart. When daily observations are unavailable to the assistant, state that coverage gap and do not infer historical changes from current lifetime totals. Read `get_finished_post` and, when needed, `get_post_history` for content, publishing and recorded provenance. Observe returned limits/cursors; a three-post snapshot cannot support an exhaustive seven-day claim. If a requested period exceeds accessible coverage, state the coverage gap.\n\nRead references/comparison-checklist.md before ranking or aggregating. Explain the time window, timezone, eligible/total post counts, platforms/accounts, metric freshness and exclusions. Distinguish creation date from public publication date; private drafts and provider-accepted inbox uploads are not measured public posts. Missing metrics remain unavailable, not zero. Compare within the same account/platform and reasonably comparable post ages; report small samples and unequal ages. Do not combine platform counters as though they measure identical audiences.\n\nInspect the strongest and weaker actual content to separate topic, hook, style and product treatment. State findings as observations, interpretations and testable hypotheses; simultaneous changes prevent attributing lift to one creative element. Report actionable next experiments and the evidence needed to judge them. Profile visits, clicks and installs require actual observed measurements/attribution; do not infer conversion from engagement.\n\nWhen asked to create from a winner, state the ranking metric/basis, read the saved post and recover its ORIGINAL recorded Favourite with `get_reference_template` and `get_reference_image`. Do not create a new Favourite from the winner or guess provenance. An explicit user-selected style overrides winner-based selection; honor the active date window. If provenance is missing, use only an already authorized fallback and label it honestly, otherwise ask for the missing choice. With insufficient comparable measurements, report that no defensible winner is established; use an authorized selected style when available. Keep visual choice separate from the new topic's research. Continue the authorized request through ../research-content/SKILL.md as needed and ../create-content/SKILL.md without repeating resolved questions.\n"
}

SHA-256 of public snapshot: d6459728ce7b7c79f7ae70d6cf394462049cbb83406002c2a17f33f91df23b95