Skills Library
Every Cowork skill DH runs on, grouped by job. What triggers it, what it touches, what it produces.
The full roster. Skills with a real SKILL.md get a real paragraph, trigger conditions and all. Skills that only ship as a compiled .skill binary get named honestly, best-guess purpose only, no invented detail.
Content pipeline
dh-content-engine
Turns any drop, a rough note, a transcript, a brain dump, a photo set, a testimonial, into a Review-ready DH Content Feed entry. Classifies into 1 of 8 content types, mines stories/people/tools into their databases, writes in Andrew’s voice stack, runs the QA gate, parks at Status=Review. Never publishes on its own.
content-capture
Quick-capture for async material that isn’t a formal meeting, DM threads, stray ideas, half-formed notes, so they land somewhere instead of evaporating. (Meeting-logger explicitly routes non-meeting async threads here instead of logging them as meetings.)
nano-banana
Generates AI images via Gemini’s Nano Banana 2 (
gemini-3.1-flash-image-preview). Triggers on any “generate an image,” “make me a graphic,” “create a transparent asset” type request. The default image generator for DH visuals, thumbnails, and brand assets.kinetic-deck
Builds interactive HTML experiences with kinetic typography, per-character animation, wave effects, and mouse-driven interactivity. For decks or landing pages that need to feel alive rather than static, animated SVGs included.
slide-deck
Builds slide decks. Best-guess from filename; has its own
references/components.md on the Mac not read for this page.prompt-intel
Generates all AI properties for Brand Builder Club prompt library entries (Writing & Thinking, Mood Board, Scene Generator, Style Transfer) in one interconnected pass. Replaces Notion AI autofill with context-aware output. Single-entry or full-database batch.
fb-event-copy
Writes Facebook Event copy for Brand Builder Club workshop promos. Triggers on “write the event copy,” “promote the workshop,” or workshop details plus a mention of Facebook.
toolbox-extract
Extracts and catalogs tools referenced across DH content into the toolbox system. Runs as a scheduled task at 4 AM (per the Tool-Cost-Map), pulling from whatever surfaced that day.
Client & meeting intelligence
The Pyramid Principle family: JSON-first scratchpad, quality gates, Stranger-Reader Test before anything ships. meeting-logger is the canonical brand-agnostic reference the rest inherit from.
| Skill | What it does |
|---|---|
| meeting-logger (v0.20.0) | Dual-writes any logged meeting to the master DH - MEETINGS DB and the project’s child Meetings DB. Builds TL;DR + Decisions-with-Rationale + DRI-owned Action Items + Open Questions + Artifacts as a JSON scratchpad, then renders to Markdown. Pushes action items to DH - TO DO LIST, saves the verbatim transcript as a child page. The canonical Pyramid Principle reference every sister skill below points back to. |
| meeting-intel (v2.0) | Processes client meeting transcripts into per-client Notion meeting databases (Heather’s Heroes, bex+Co., Jack Lind, Landon Smith, etc). Builds client context first (business, plan type, recent meeting history) so pain points get tagged recurring-vs-new and action items get checked for follow-through. Writes 6 properties: Speaker Map, Client Pain Points, Host Strategic Moves, What Clients Say, Universal Insights, Action Items. |
| 22social-intel (v2.0) | Same JSON-first pipeline, scoped to the 22Social project orbit (Rene, Gary, Bex, franchise/stylist tooling). Dual-writes to 💡 22Social Ideas (one page per session, full body template) and 22Social - Project Log (only genuine strategic pivots, never tactical tasks). |
| training-intel (v2.0) | Enriches existing General Training Library entries: generates 6 AI properties (Deep Training Summary, Content Repurpose Kit, Sales & Proof Extraction, Teaching Gold, ICA Resonance Map, Shelf Life Assessment) from the page’s transcript. Enrichment only, doesn’t create new entries, single or full-database batch. |
| video-intel (v2.0) | Same enrichment pattern for the Public Videos database: 7 AI properties (Deep Video Summary, Mini-Training Breakdown, SEO & Discovery Pack, Content Repurpose Machine, AHA Moment Index, Offer Bridge, full AI Description). Fetches transcript via youtube-transcript if only a YouTube link is present. |
| workshop-replay (v2.0) | The CREATE counterpart to training-intel: builds a brand-new, fully-populated General Training Library page from a raw workshop transcript, all 14 properties plus the 6 deep-analysis properties, verbatim transcript saved as a child page. |
YouTube content engine
morning-briefing
Daily 7 AM scheduled brief: YouTube Studio scan, overnight comments, search trend deltas on core topics, competitor uploads, evergreen performance shifts. Synthesizes into a 1-page brief with today’s recommended action and 3-5 comment questions worth answering on stream. Logs to the YouTube Growth System.
post-stream
Processes a finished live into the full next-day package: stream summary, top moments, 3-5 Shorts cut-list with hooks and thumbnail prompts, ranked recap titles, SEO description, cross-platform posts, tomorrow’s hook. Run after every stream, no exceptions.
clip-factory
Takes a stream URL plus post-stream’s cut-list and outputs 3-5 ready-to-upload Shorts: vertical 9 face-tracked reframe, burned Poppins-brand captions, hook overlay on the first 2 seconds, Nano Banana thumbnails, full upload metadata. The editing automation that removes the daily engine’s biggest manual time sink.
stream-prep
Prepares for an upcoming live stream. Best-guess from filename; likely the pre-stream counterpart to post-stream (talking points, hook, agenda).
watch
References Groq/Whisper as a transcription fallback (per the Tool-Cost-Map), suggesting it watches/monitors video or stream content and transcribes when captions aren’t available. Best-guess beyond that.
Deploy & sync
| Skill | What it does |
|---|---|
| bbb-deploy (v2.1) | The full Brand Builder Bot plugin ship pipeline: version bump across 4 targets (plugin.json, CLAUDE.md, README, every skill header), brand-agnostic audit (halts on hardcoded DB names), plugin schema check (halts on Claude Desktop-incompatible fields), package, push to GitHub, deploy to dh-share, update the hub + download page, Notion backups for changed files only, bbb-knowledge-sync handoff, and a CURRENT.md truth-pointer write. 4-surface version parity is the final gate. |
| dh-share-deploy (v3, Apr 20 build) | Deploys HTML files and deliverables to share.designhacker.com via the GitHub auto-deploy pipeline. Default publishing step for any web-viewable DH output, client or internal. |
| brand-library-sync | Dual-writes important session knowledge (SOPs, API gotchas, workspace rules, architectural decisions) into the DH Brand Library Notion database. Fires on “save this,” “back this up,” or proactively at the end of any session where something durable emerged. |
| bexco-update-log | Background logger: every meaningful change made inside the Bex+Co Notion workspace during a session gets a row in the Bex+Co - Update Log database. Runs alongside other work, not something asked for explicitly. |
| session-sync-log | The meta-record skill every other skill above calls at the end of its run (“log to the Activity Log via session-sync-log”) so other Cowork sessions and devices stay in sync on what just happened. |
Workspace & ops
command-center-setup
Provisions the master Command Center Notion structure (per meeting-logger’s error handling: creates the
DH - MEETINGS master DB if missing). The one-time scaffolding skill other skills fall back to when they can’t find their write target.project-page-builder
Sets up a project-type-aware page in Active Projects: Course, Client Engagement, Launch, Content Series, Internal SOP, or Product Build, each provisioning the right child databases. Product Content is the spine. Guided ingestion follow-up for Course/Client/Series/SOP/Product types.
brand-snapshot-builder
Hands off into project-page-builder post-setup (per that skill’s own trigger list), so this one runs earlier, capturing a brand/client snapshot during intake. Best-guess beyond that cross-reference.
x-intel
X/Twitter intelligence, powered by Apify per the Tool-Cost-Map’s scraping line (“Apify: lead-finder, x-intel scraping”). Best-guess on the rest.