The problem
The README describes the skills as tools “for promoting and syndicating your work on social media”. It frames the design as a closed loop: “every post is tagged at compose time so the analytics skills can attribute engagement back to the format that produced it”. The loop then auto-generates skill-config recommendations from each week’s data. In his LinkedIn post he calls the result “my weekly growth-ops stack”.
The blueprint (as he describes it)
- Tag every post at compose time. Each compose skill attaches a format tag (format:verbatim-quote, format:teaser, format:carousel, format:link-share and others). The README sums up the loop as “Compose with format tag → adversarial review → user publish → Buffer fans out → Measure attributes engagement back to format → Recommendations feed next compose run.”
- Turn one newsletter post into three kinds of social post.
- /promote-newsletter pulls verbatim snippets from a beehiiv post and only trims them to fit character limits.
- /tease-newsletter writes short teaser hooks per channel “without spoiling the punchline.” The README recommends it as the default for LinkedIn “(verbatim quotes underperform there)”.
- /carousel-newsletter makes a 10-slide illustrated carousel for Instagram, LinkedIn, Facebook and Threads with Gemini 2.5 Flash Image, using his brand banner as the style reference. Quote slides stay verbatim.
- Ground copy in his own voice. A shared helper builds a voice corpus from his recent beehiiv newsletters (written register) and YouTube livestream transcripts (spoken register).
- Keep one CTA so the DM automation keeps working. A shared script generates the same Comment “newsletter” CTA every time. It is a Manychat trigger word, and the README says not to edit it ad hoc.
- Review before anything queues. A fresh agent reviews each draft against the source and the rules. Failing drafts go back to compose. Passing drafts go to him. A helper then enforces character limits, skips dead or below-threshold channels, attaches the tag and checks the queue for overlap.
- Promote the rest of the work. /promote-github turns merged PRs, commits and releases into posts with one hero image per theme, and defaults to instant-publish. /crosspost-newsletter syndicates the full article to LinkedIn, Substack and Medium with the canonical URL pointing home. A short Buffer companion post keeps the LinkedIn article’s engagement attributable.
- Measure each surface. /buffer-stats reads Buffer’s data and analytics. /linkedin-stats scrapes LinkedIn analytics for subscribers, followers and per-post engagement. In his LinkedIn post he says it degrades gracefully when LinkedIn changes its analytics pages: “screenshot + handoff instead of silent stale numbers”. TikTok Studio and Threads Insights scrapes run only interactively, in his logged-in Chrome.
- Roll it up weekly with /flywheel. It is a “Cross-platform weekly rollup keyed to your 5 growth priorities,” combining Buffer, LinkedIn, YouTube and beehiiv into one report with per-channel ROI scoring “to surface deprioritization candidates.” His LinkedIn post describes “Actual vs target vs status (green/yellow/red) per priority.” It grades last week’s hypotheses and writes next week’s into a ledger.
- Keep the queue clean. /audit-buffer-queue flags bunching (gap < 3h between posts on the same channel), theme over-saturation, untagged posts and dead channels, and recommends cancel, reschedule or tag actions. /tune-posting-schedule changes the time slots “so bunches stop recurring,” only after he approves each channel.
- Run it on Sunday, decide on Monday. An optional cron runs the full review headlessly every Sunday at 09:30 local time. It is “Read-mostly” and does not auto-apply destructive queue actions. It surfaces them for his Monday review.
Stated results
None stated beyond cost and time per deck.
- Carousel: ”~$0.40 per deck, ~15 min wall-clock.”
- The README also gives ”~$0.04/image” for each GitHub hero image.
- No subscriber, follower, reach or engagement figures are stated in either source.
Caveats
- Mike Lady runs Enterprise Vibe Code, whose site tagline is “Dependable workflows for ecommerce teams.” That suggests he sells workflow work to ecommerce teams Unverified · : no offer or pricing checked. The repo is free and public.
- The README describes the design and recommended use. It does not show outcomes. The LinkedIn recommendation for teasers over verbatim quotes is his statement with no data shown.
- Several measurement skills scrape logged-in browser sessions. The TikTok and Threads scrapes are interactive only and do not run unattended.
- /promote-github defaults to instant-publish rather than queueing.
- Setup is technical. It needs Claude Code, built Go helpers, the gstack browse component and Google Cloud authentication for image generation.
- Handles: LinkedIn linkedin.com/in/mikelady, YouTube @EnterpriseVibeCode, GitHub michaellady (all verified 2026-10-23: linked from enterprisevibecode.com). No X, Bluesky, Threads, Instagram, TikTok or Mastodon profile could be verified (checked 2026-10-23).
Sources
- https://github.com/michaellady/claude-social-media-skills (primary: his public GitHub repo README)
- https://www.linkedin.com/posts/mikelady_shipped-my-weekly-growth-ops-stack-this-weekend-activity-7452147091809800192-de6W (his LinkedIn post, “Shipped my weekly growth-ops stack this weekend”)
Want agents like these working your distribution? See the Agent Stack Teardown. Get new teardowns weekly: subscribe (double opt-in).
Get one teardown a week.
Same format, a new operator every issue. Double opt-in.