Antwork analytics
Agent Skills for the Antwork MCP connector
npx -y skills add iker-gonzalez/antwork-skills --skill antwork-analyticsAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Use when the user wants to understand how their social content is performing through Antwork — pulling engagement metrics, ranking top posts, spotting per-platform trends, comparing time periods, or turning numbers into a what-to-do-next report. Trigger on phrases like "how are my posts doing", "what's my best post", "engagement report", "analytics", "which platform is working", "show my growth", "refresh my metrics", or any request for social performance data.
SKILL.md
3.6 KB, as published. Nobody here has run it
Reporting on performance in Antwork
Antwork's analytics tools return BigQuery-style tabular data — { schema: { fields }, rows, rowCount } — not charts. Your job is to pull the right table, then synthesize it into a decision: what's working, on which platform, and what to make more of. Numbers without a recommendation are not a report.
1. Pick the right query — don't pull everything
| Question | Tool | Notes |
|---|---|---|
| "What are my best posts?" | get_performance(limit) | Lifetime totals per published post (likes, comments, shares, impressions, totalEngagement). Best for rankings. 1–50, default 30. |
| "How is engagement trending?" | get_engagement_history(days) | Daily series per platform + an "all" aggregate, plus platformTotals/rangeTotals. Best for trends over time. 1–365 days, default 30. |
| "How did this post grow?" | get_post_history(post_id, days?) | Per-day metrics for one post, both cumulative and daily-delta. Best for single-post lifecycle. Omit days for full history. |
| "When should I post?" | get_optimal_posting_times() | The workspace's configured times + timezone + selected accounts. |
Start with get_performance for almost any "how am I doing" ask, then drill in with the others.
2. Don't burn live quota by reflex
refresh_post_metrics(post_ids) hits the platform APIs live, right now — ~1–2s per post, max 25, and it spends real platform rate-limit quota. The stored numbers are already refreshed by a nightly cron, so they're at most a day stale. Only call refresh_post_metrics when the user explicitly needs up-to-the-minute numbers (e.g. a post that just went viral this morning), and cap it to the handful of posts that matter.
3. Read the table, then interpret
The rows are raw. Turn them into findings:
- Top performers — rank by
totalEngagement, but look past the number to the pattern. What do the top 3 share — format, hook style, topic, length, platform, posting time? - Per-platform trends — from
get_engagement_historyplatformTotals: which platform is growing, which is flat, where's the effort/return mismatch. - Outliers — one post 5× the median is a signal, not noise. Name what made it land.
- Gaps — platforms with no recent posts, or with traffic but no posting cadence.
4. Tie every finding to an action
The report's value is the "so what." Each finding gets a next step, and the strongest ones feed straight into other skills:
- Winning hooks/topics → hand to
antwork-ideasto generate more in the same vein. - Best posting windows → feed
antwork-calendarwhen scheduling. - A high performer worth a sequel →
antwork-repurposeit to other platforms.
5. Output
For a quick ask, give a tight terminal summary (top 3 posts + the one thing to do more of). For a full report, write a markdown file following templates/analytics-report.md — header with workspace/date/range, executive summary, top-performers table, per-platform trends, and a prioritized recommendations list. Always state the date range the numbers cover, so a stale report can't be mistaken for current.