Update to Ravenpost

{
  "description": "Read a Ravenpost workspace's social analytics — follower counts per connected account, a 30-day follower series, and engagement on recent posts — and answer \"when should I post?\" without inventing a measured hour. Use when the user asks how a post or an account is doing, how many followers they have, which post did best, or when they should publish (\"how did last week do?\", \"what's the best time to post on Instagram?\"). Covers the difference between a metric that is zero and one that was never collected, which is the mistake this data invites.",
  "included_files": [],
  "name": "measure-and-time-posts",
  "skill_md_contents": "---\nname: measure-and-time-posts\ndescription: Read a Ravenpost workspace's social analytics — follower counts per connected account, a 30-day follower series, and engagement on recent posts — and answer \"when should I post?\" without inventing a measured hour. Use when the user asks how a post or an account is doing, how many followers they have, which post did best, or when they should publish (\"how did last week do?\", \"what's the best time to post on Instagram?\"). Covers the difference between a metric that is zero and one that was never collected, which is the mistake this data invites.\n---\n\n# Measure posts, and time the next one honestly\n\nOne tool does the measuring, and it is read-only:\n\n- `get_analytics` — followers per connected account with a 30-day daily series,\n  plus engagement on recent posts.\n\n**Settle the workspace first** (`list_workspaces`); analytics are per workspace,\nand an agency's clients are measured separately for a reason. Full argument list:\n[`../publish-a-social-post/references/tools.md`](../publish-a-social-post/references/tools.md).\n\n## Absent is not zero\n\nThis is the rule that decides whether your answer is true.\n\n- An account's `followers` is **`null`** when that platform reports none — not\n  when the account has none. Which platforms report nothing today is named in\n  `get_analytics`'s own description; trust that over memory, because it moves\n  as permissions are granted. Say \"Instagram reports followers; LinkedIn\n  doesn't expose them\", never \"LinkedIn: 0\".\n- A post metric that is **missing** was not collected. Do not add it into a total\n  as a zero, and do not present a sum as complete when part of it is unknown.\n- `collectedAt` says when the numbers were last read from the platforms. If the\n  user is surprised by a stale number, that is the explanation.\n- A post's `source` is `ravenpost` for one published from here and `platform`\n  for one that was already on the account and imported so its engagement counts.\n  Both are real posts; only the first went out through Ravenpost.\n\n## When to post\n\nRavenpost does not compute a \"best time to post\", and nothing it returns is\none. When the user asks:\n\n- **If the workspace runs a posting queue** (`list_queue_slots`), its slots are\n  the times this workspace has chosen to post. `create_post` with\n  `action: \"queue\"` takes the next free one, with no arithmetic.\n- **Otherwise ask for a time**, or offer one plainly as a suggestion — never as\n  a finding.\n- You may point at which recent posts earned the most engagement in\n  `get_analytics`, but say how many posts that rests on. A handful of posts is\n  an anecdote, not a measurement, and it does not make an hour \"the best\".\n\nWhen you do schedule, pass `scheduledAt` as an absolute ISO 8601 datetime **with\nthe offset**, worked out in the workspace's timezone (`list_workspaces` gives\nit), and show the user the local time you computed — a timezone mistake should\nbe visible before the post is scheduled, not after it goes out.\n\n## Never\n\n- Never invent a benchmark (\"good engagement for this size is ~3%\"). Report what\n  was measured.\n- Never present an hour as a measured best time to post. Nothing here measures\n  one.\n"
}