Skill instructions
content-calendar-audit1.94 KB
View saved version →
---
name: content-calendar-audit
description: 'Check what is planned in a Planable workspace for a week or a month and spot empty days, undated drafts, unapproved posts and publishing errors. Use when the user asks to review, audit or check their content calendar.'
---
# Content calendar audit
Review what is planned for a period and point out the gaps and risks.
## 1. Scope
- Workspace. Ask if it is not clear.
- Period. Default: the current week. Accept next week, this month or a date range.
- Pages. Default: all pages in the workspace.
## 2. Fetch
- If show_calendar is available, call it for the period (view week or month). It shows the calendar card and returns the posts grouped by day in the workspace timezone.
- Otherwise call list_posts with scheduledAfter and scheduledBefore set to the period, and page through all results.
Then add:
- list_posts with postStatus NO_DATE_SET: drafts not placed on the calendar yet.
- list_posts with postStatus WITH_ERRORS: posts whose publishing failed.
- list_posts with approvalStatus NOT_APPROVED and PARTIAL_APPROVED, limited to the period: posts that may go out without full approval.
## 3. Audit
Do not repeat the posts the calendar card already shows. Summarize instead:
- How many posts are planned, on which platforms, and how many are confirmed for publishing (status scheduled) rather than only dated.
- Days with nothing planned.
- Posts with no date.
- Posts with publishing errors, each with its page and a preview.
- Dated posts that are not fully approved, marked urgent when they are due within 48 hours.
Then give 2 to 4 specific recommendations from this data, such as moving an undated draft to an empty day. No generic best-practice advice.
Say which timezone the dates are in: list_posts returns UTC, show_calendar uses the workspace timezone.
## 4. Follow up
Offer to draft posts for the empty days (the draft-post-batch skill), to go through the approvals, or to audit another period or workspace.
content-pattern-intelligence2.46 KB
View saved version →
---
name: content-pattern-intelligence
description: 'Find what the best posts in a Planable workspace have in common and suggest 3 ideas to test in the next batch. Use when the user asks what content works, why their top posts perform, or what to post more of.'
---
# Content patterns
Find what the best posts have in common and turn it into ideas to test next.
## 1. Scope
- Workspace and pages. Default: all pages in the workspace.
- Period. Default: the last 90 days. Patterns need about 15 posts with metrics or more. With fewer, say the findings are tentative.
## 2. Fetch
- list_pages, then get_page_metrics for the period to see the posts per page and metricsFetchedAt.
- If the data is more than a day old, offer refresh_post_metrics. It only queues a sync that takes a few minutes, so offer to continue with the stored data.
- get_post_metrics for the period, paging through all posts. If a call is slow or fails, fetch one page at a time over shorter date ranges. If a range keeps failing, skip it and mention it in the data notes.
- textPreview is short. Open the top posts with get_post when you need their full text.
## 3. Rank three ways
- Top 10 by engagement (likes, comments, shares).
- Top 10 by impressions or views.
- Top 10 by engagement rate (engagement divided by impressions). It shows content quality apart from reach.
Use all posts when there are fewer than 10, and keep the bottom posts for comparison.
## 4. Read the posts
For each top list, read the post text and look for what repeats:
- Format: length, lists, line breaks, emoji, a question or call to action at the end.
- Hook: the first line (question, bold claim, number, story, speaking to the reader).
- Topic: tips, behind the scenes, opinion, product, social proof, data.
- Tone: personal or polished, opinionated or neutral.
- Timing: days or times where the dates cluster.
Then compare with the bottom posts. A pattern counts only when the top posts have it and the bottom posts mostly do not.
## 5. Write it up
Start with a short summary: what is working (2 or 3 lines), 3 ideas to test next (one specific sentence each) and one thing to do less of.
Then, per ranking, 2 to 4 patterns, each with: what it is, one real example (the first words of a top post), why it likely works (honest, no overclaiming) and a concrete test for the next batch.
End with what is not working and data notes: posts analyzed, period, pages and any skipped ranges.
Offer to draft a batch that tests the ideas (the draft-post-batch skill).
cross-client-metrics-overview1.76 KB
View saved version →
---
name: cross-client-metrics-overview
description: "Rank the user's Planable workspaces by engagement rate, impressions or posts, and flag the ones that need attention. Use when the user wants to compare clients or brands, or asks which accounts are doing best or worst."
---
# Cross-client comparison
Compare performance across several workspaces in one view.
## 1. Scope
- Workspaces. Default: all of the user's workspaces. With more than 15, say how many and ask whether to narrow it down.
- Period. Default: the previous calendar month.
- Ranking metric. Default: engagement rate. Alternatives: impressions, engagement, posts published, audience.
For a single workspace, use the monthly-performance-summary skill instead.
## 2. Fetch, per workspace
- list_pages for the page ids.
- get_page_metrics_summary for audience, impressions, engagement, engagement rate and posts.
- get_post_metrics_summary for the post totals.
- Optionally the same two summaries for the previous period, for the change against it.
If a workspace fails, for example because it has no analytics add-on, mark it "No data" with the reason and carry on with the others.
## 3. Present
1. A ranked table: workspace, the ranking metric, impressions, engagement and posts, with a totals row. Keep workspaces without data in the table as "No data".
2. 3 to 5 callouts, only where the data supports them: top performer, most improved, needs attention (engagement dropping or very low) and low activity (fewer than 3 posts in the period).
3. Notes: pages without metrics (unsupportedPages), X left out of the page totals, and workspaces with very different amounts of data.
Be direct about the workspaces that underperform. Round the numbers and show each one once. Offer a full report for any single workspace on request.
draft-post-batch2.68 KB
View saved version →
---
name: draft-post-batch
description: 'Write several social posts from one brief and save them as drafts in a Planable workspace, ready for review. Use when the user wants a batch of posts, a week of content, or the same idea adapted for several platforms.'
---
# Draft a batch of posts
Write several social posts from one brief and save them in Planable as drafts. Nothing is published.
## 1. Gather the brief
Confirm before writing:
- Workspace (client or brand). If it is not named, find it with list_workspaces and let the user pick. Do not guess.
- Pages or platforms to post on.
- Topic or brief.
- Number of posts. Default 5.
- Format where the platform has several: post, story or reel (classification).
- Tone or angle. Optional. Without one, use the platform's natural register.
- Dates. Optional. If the user wants dates but did not pick them, call get_posting_times for those pages and spread the posts over the suggested slots.
- Labels or a campaign. Optional. Ask only if the user mentions them, then look them up with list_labels or list_campaigns.
## 2. Find the pages
Call list_pages for the workspace and match the requested platforms. If a platform has several pages, ask which one.
## 3. Write all drafts first
- Vary the hook and structure across the batch. Do not repeat one template.
- Fit each platform: X 280 characters; LinkedIn up to about 3,000 with line breaks; Instagram and TikTok up to about 2,200, caption style, hashtags at the end; Facebook short and conversational; YouTube descriptions informative.
- Active voice, short sentences, contractions. Avoid filler words such as seamless, streamline, effortlessly, unleash and all-in-one.
- Instagram, TikTok and YouTube posts need media. Without media, say the draft will need some before it can publish, or offer another platform.
Show all drafts as a numbered preview and wait for the user's go-ahead, unless they already said to create them directly.
## 4. Save them in Planable
- Same text on several pages: create_grouped_post.
- Different text per platform: create_post once per page.
- Create the posts one call at a time, so each validation error shows next to its post.
- Pass scheduledAt as a proposed date only. Set publishAtScheduledDate only when the user explicitly asked to schedule the posts for publishing.
- Media: public URLs go in mediaUrls, media library items in mediaIds. For files attached in the chat, use the upload tools.
## 5. Report
Say how many drafts were created, in which workspace and on which pages, with their dates. List any validation errors next to the post they belong to. Remind the user that the posts stay drafts until someone approves or schedules them, and offer to change any of them with update_post.
monthly-performance-summary2.28 KB
View saved version →
---
name: monthly-performance-summary
description: "Turn one Planable workspace's analytics into a client-ready report with key numbers, platforms, top posts and takeaways. Use when the user asks for a monthly report, a performance summary, or how a client's social media did last month."
---
# Monthly performance report
A client-ready summary of one workspace's social performance for a period.
## 1. Scope
- Workspace. Ask if it is not clear.
- Period. Default: the previous calendar month. Point it out when the period is a partial month.
- Pages. Default: all pages in the workspace.
Analytics needs the analytics add-on (hasAnalytics from get_workspace_details) and the user's analytics permission. If a tool rejects for that reason, say so plainly and stop.
## 2. Fetch
- list_pages for the page ids.
- get_page_metrics for the numbers per page (audience, posts, impressions, engagement, engagement rate and the platform's own metrics).
- get_page_metrics_summary for the totals.
- get_post_metrics_summary for the post totals in the period.
- get_post_metrics for the top posts. Fetch enough to rank the top 5 by engagement.
- For the change against the period before, repeat the two summary calls for the previous period of the same length.
If metricsFetchedAt is old, offer refresh_page_metrics and refresh_post_metrics. They only queue a sync, and newer data lands after a few minutes, so offer to continue with the stored numbers.
## 3. Report
1. Key numbers: audience, impressions, engagement, engagement rate and posts published, each with the change against the previous period when you have it.
2. By platform: one line per page or platform.
3. Top 5 posts: page, a short preview and the number that put it there.
4. What worked: 2 or 3 specific points from the data.
5. What to watch: 1 or 2 specific points.
6. Notes: pages without data (unsupportedPages), how fresh the data is (metricsFetchedAt), and that X figures cover only the posts published through Planable.
Round the numbers (12K, 4.2%) and use plain language a client understands. If the month was quiet, say so, and never invent a trend.
If you can create documents or files here, offer a shareable version of the report. Otherwise keep it in the chat.
If the user uploads an analytics export instead, build the same report from the file.
pending-approvals-roundup1.76 KB
View saved version →
---
name: pending-approvals-roundup
description: 'List every Planable post waiting for approval across workspaces, with the ones going live soon first. Use when the user asks what needs their approval, what is stuck in review, or what is holding up the calendar.'
---
# Approvals roundup
One overview of the posts that wait for approval across workspaces, most urgent first.
## 1. Scope
- Workspaces: one, a few, or all. Default: all of the user's workspaces. With more than 10, say how many you will check and ask whether to narrow it down first.
- Whose approval: the user's own (approvalStatus PENDING_MY_APPROVAL, the default) or every post that is not fully approved (NOT_APPROVED and PARTIAL_APPROVED). REQUESTED_BY_ME shows the posts the user sent to others for approval.
## 2. Fetch
For each workspace in scope, call list_posts with the approvalStatus filter and page through until nextOffset is empty. Also call list_posts with postStatus WITH_ERRORS for the same workspaces, since failed posts often block the plan too.
## 3. Present
Group by workspace. For each post show the page, a short text preview, its date and its approval state.
- First the posts dated within the next 48 hours, marked as urgent.
- Then the other dated posts by date, then the posts with no date.
- Publishing errors in a separate block.
If nothing waits for approval, say so in one line.
## 4. Act on request
Offer to open a post in full (get_post), approve it (approve_post), or send it back with a reason (disapprove_post, then create_comment with the reason). Approve or reject only the posts the user names, after they confirm, and never by default. Approving needs an approval workflow in the workspace (approvalWorkflow from get_workspace_details is not "none") and the user's approval permission.