claude-growth-engine

User-invoked skill to upload one or more local Markdown files to Travis's Google Drive as native Google Docs (not as .md attachments) using the gws CLI. Default destination is the "Claude-Uploads" folder at Drive root. Trigger phrases include "upload to drive", "convert to google doc", "send this to claude-uploads", "publish these to drive", "drop these markdown files into drive", "make google docs from these". For batches of 2+ files, dispatch one parallel subagent per file.

Edit a LinkedIn draft before its copy lands in the Jira CSV. Two quick passes (hook/tightness, then a hard compliance gate) plus a scaled score. Enforces linkedin-writing-guide.md so voice/policy rules are checked by a fresh pass, not self-graded by the drafter. Use inside linkedin-process.md (Step 4.5) or standalone on any drafts/ file.

When the user wants webinar promotional emails — the 3-email sequence (intro at T-10d, value reiteration at T-3d, last reminder at T-1d) for both Cerkl and the partner, plus the 1:1 sales outreach snippet reps paste into active deals. Trigger phrases include "promo emails", "webinar emails", "email sequence", "promotional emails", "draft the emails", "T-10 email", "day-before email", "last reminder email", "sales outreach snippet", "sales email template", "something for sales to send", "1:1 rep outreach". Run this AFTER the brief is filled out.

Sub-agent brief for the Deal Report's inference pass — reads ONE flagged deal's cleaned email threads + structural metrics and returns sentiment, the open next-step, and a short risk narrative. Invoked only on deals the rules band watch/at_risk (or a deal Travis names). Never sets the band; explains it.

When the user wants to convert a rendered HTML file to a PDF, when a process is generating a print-format deliverable (one-pager, slide handout, designed report), when the final step of a render pipeline needs to produce a .pdf sibling. Triggers on phrases like "render to PDF," "save as PDF," "convert to PDF," "print this." Built-in verify gate: runs html-overflow-detector first; if the HTML has layout overflow, the PDF is NOT written and the caller is told to apply remediation. Invoked automatically by one-pager-process.md as the final step after md-to-html.

Render the full Canva asset pack for one webinar by dispatching one parallel sub-agent per manifest in the event's canva-manifests/ folder. Each sub-agent invokes template-fill against its assigned manifest and returns the Canva edit URL. The orchestrator appends each URL to its manifest YAML (primary source-of-truth) and mirrors the URL into the Drive MAP's Deliverable column (bonus for the internal team). Trigger phrases: 'render the canva assets for [webinar]', 'kick off canva render for rachel', 'run the asset pack render', 'fill all the webinar templates'. Inputs: event-folder path + headshot Canva asset ID. Output: chat-printed roll-up of role → URL + warnings + manual_drag_required list.

Generate one on-brand Canva asset from a Cerkl brand template by replacing text, inserting/swapping image fills by existing asset ID, and optionally deleting unused slots (e.g. drop speaker 2 on a solo speaker card) — the proven Path B loop (create-design-from-brand-template → start-editing-transaction → perform-editing-operations → commit-editing-transaction). Trigger phrases: 'render this Canva manifest', 'build the speaker card', 'generate the LinkedIn stat card from template X', 'fill this template with these values', 'run the template-fill skill'. Atomic and channel-agnostic — webinar / LinkedIn / blog / video processes all invoke it (typically dispatching one sub-agent per template for parallel rendering). Input: a manifest (template_id + page_index + text values + image asset IDs + optional delete-slots + commit mode). Output: {design_id, edit_url, view_url, warnings[], manual_drag_required[]} — the Canva URL is the deliverable handed to whoever finishes the asset (today: Furqan). Skill is fully driven

Use after `icpro-blog-editing` finishes a post — takes the `_live.md`, uploads it to Google Drive as a Doc with the `ICP` naming segment, then inserts the Drive URL into the target week's Jira CSV row matching the synthesized slug. This is the final step before Travis imports the CSV to Jira. Triggers on phrases like "publish the [slug] ICP post", "run publishing on the [slug] ICPro draft", "upload [slug] to Drive and wire into Jira CSV". Prerequisite: a `_live.md` in `blog-posts-live/` AND a Jira CSV scaffold at `content-plan/jira/imports/YYYY-Www.csv` for the target week. Output: Drive Doc URL + updated CSV row.

Use after `seo-blog-editing` finishes a post — takes the `_live.md`, uploads it to Google Drive as a Doc, then inserts the Drive URL into the target week's Jira CSV row matching the brief slug. This is the final step before Travis imports the CSV to Jira. Triggers on phrases like "publish the [slug] post", "run publishing on the [slug] draft", "upload [slug] to Drive and wire into Jira CSV". Prerequisite: a `_live.md` in `blog-posts-live/` AND a Jira CSV scaffold at `content-plan/jira/imports/YYYY-Www.csv` for the target week. Output: Drive Doc URL + updated CSV row.

When the user wants to build a new process, set up a new workflow, formalize a recurring task, spin up a new agent/researcher/role, or turn an ad-hoc job into something Claude can run repeatably. Triggers on phrases like "build a new process," "set up a new workflow," "create a new job for X," "spin up a [role]," "make this repeatable," "formalize this." Outputs a `[job]-process.md` orchestrator, any new `SKILL.md` files needed, and routing-table updates in every parent CLAUDE.md on the path. Domain-agnostic — works for marketing, sales, ops, research, personal-assistant work, anywhere in this workspace.

Use when drafting a Cerkl SEO blog post from a completed pre-writing file — produces the full markdown draft (body + FAQ + FAQ schema) ready for the editing pass. The editing pass owns CTA-marker placement; drafting only owns the prose. Triggers on phrases like "draft the blog post for [slug]", "write the [topic] post", "turn this pre-writing into a draft", "draft this blog". Prerequisite: a pre-writing file at `blog-posts-pre-writing/`. Output: `blog-posts-draft/YYYY-MM-DD_[slug]_draft.md`.

When the user wants to set up a new webinar project, kick off a webinar, scaffold a new event folder, or start planning a new webinar with a partner. Trigger phrases include "new webinar", "set up a webinar", "kickoff webinar", "start webinar project", "scaffold webinar folder", "create webinar project". Run this AFTER the topic and date have been agreed with the partner.

Use when preparing a Cerkl SEO blog post for drafting. Reads the upstream SEO brief from `seo/briefs/<slug>.md`, carries its schema decisions (pillar, solution, keywords, internal links) into a Webflow-ready pre-writing file, and adds the per-post decisions the brief doesn't cover (Top/Middle/Bottom CTA variants, meta description, full H1/H2/H3 outline). Triggers on phrases like "pre-writing for [topic]", "set up the blog brief for [post]", "outline this blog post". Output: `blog-posts-pre-writing/YYYY-MM-DD_[slug]_pre-writing.md`.

Use when finalizing an Internal Comms Pro (internalcommspro.com) blog draft — runs a two-pass editing process (structural edit, sentence-level rewrite, line edit, score) to remove AI writing patterns and tighten prose. Produces a publication-ready `_live.md`; Drive upload and Jira CSV handoff happen in the separate `icpro-blog-publishing` skill. Triggers on phrases like "edit the [slug] ICP draft", "run editing on this ICPro post", "finalize this ICP draft", "score this ICP draft". Prerequisite: a draft at `blog-posts-draft/`. Output: `blog-posts-live/YYYY-MM-DD_[slug]_live.md`.

When the user wants to verify an HTML page renders without layout problems before generating a PDF, when a rendered PDF has visible overlap or cut-off content, when a render-verify-remediate loop needs a deterministic "does the layout fit?" check. Triggers on phrases like "check for overflow," "verify the layout," "did anything get cut off," "find the overlap." Invoked automatically by one-pager-process.md and any process that renders fixed-format artifacts (letter PDFs, slides, ad creatives) from variable-length content. Returns a JSON report listing overflow type, page, selector, pixel overrun, and a text snippet — plus an exit code (0 = PASS, 1 = FAIL).

When the user wants the post-event follow-up email that distributes the recording and assets to registrants. Trigger phrases include "follow-up email", "post-event email", "recording email", "send the recording", "webinar follow-up", "thank you email", "post-webinar email". Run this AFTER the live event has happened and the recording is available.

Scan HubSpot marketing emails for clone-and-edit mismatches between name, subject, body, and download links. Produces a findings report. Read-only — does not modify emails. For making changes, use draft-marketing-email.

Create a new HubSpot marketing email draft, or update copy on an existing one — subject, body, link text. Stages the change as a draft via the API; the user publishes from the HubSpot UI. Use for routine email-writing work where Claude generates or edits the content.

Run a comprehensive HubSpot CRM database audit. Analyzes contacts, companies, deals, engagement, data quality, and deliverability. Use when starting a CRM cleanup, onboarding a new client, or performing quarterly health checks.

When the user wants to brainstorm, generate, or draft new YouTube ad hook ideas — the 5–10 second openers that swap into the polished paid YouTube ad. Trigger phrases include "generate hooks", "new hook batch", "20 youtube hooks", "brainstorm video hooks", "new paid youtube hooks", "draft hooks for the youtube ad". Always produces 20 ideas in a dated batch folder, seeded by ICP pain points, Broadcast positioning, and the proven-hooks library.

When the user wants to generate Google Flow video prompts from a hook storyboard, write Flow prompts, or convert a shot list into AI video generation prompts. Trigger phrases include "write flow prompts", "generate google flow prompts", "scene prompts", "flow prompts for this hook", "prompts for the storyboard". Run AFTER `paid-youtube-hook-storyboard`. Output goes into the same winner folder, ready to paste into Google Flow video studio.

When the user wants to pick winners from a YouTube hook batch, narrow down hook ideas, or capture which hooks to move into production. Trigger phrases include "pick hook winners", "select youtube hooks", "review hook batch", "narrow down the hooks", "these are the winners". Run AFTER `paid-youtube-hook-batch` has produced an `ideas.md` file. Travis picks by hook ID; this skill captures the selection plus reasoning in `winners.md`.

When the user wants to storyboard a selected YouTube hook, turn a winning hook into a shot list, or plan camera/visual direction for a 5–10s video ad opener. Trigger phrases include "storyboard this hook", "create youtube storyboard", "shot list for the hook", "storyboard the winner". Run AFTER `paid-youtube-hook-select` has produced a `winners.md`. Run once per winner.

User-invoked skill to process raw post-event artifacts (Google Drive transcript .vtt, Zoom chat .txt, slide deck .pptx) into cleaned source-of-truth files that all downstream post-event skills (followup email, recap clips, blog recap, LinkedIn pulls) consume. NOT auto-triggered — Travis runs this explicitly after the live event, once raw files are saved into the event folder's `raw/` subfolder. Trigger phrases (only when explicitly invoked): "ingest the webinar", "process the raw files", "clean up the transcript", "run webinar ingest".

Produce a standalone, SEO-tuned blog post on a topic from the webinar that lives in the SEO blog channel archive — not framed as a "recap." The blog stands on its own as a topic-led piece, references the webinar recording as a CTA inside the body, and pulls quotes from the speakers. Trigger phrases include "recap blog", "post-event blog", "blog from the webinar", "webinar recap blog", "topic blog from the webinar". Run this AFTER `webinar-ingest` has produced cleaned source-of-truth files.

Use when drafting an Internal Comms Pro (internalcommspro.com) blog post from a completed pre-writing file — produces the full markdown draft for Wix paste, with the more authoritative ICP voice and no Cerkl-product CTAs. Triggers on phrases like "draft the ICP post for [slug]", "write the [topic] ICPro post", "turn this pre-writing into an ICPro draft". Prerequisite: a pre-writing file at `blog-posts-pre-writing/`. Output: `blog-posts-draft/YYYY-MM-DD_[slug]_draft.md`.

Use when preparing an Internal Comms Pro (internalcommspro.com) blog post for drafting — gather the topic, fill out the Wix publishing properties (title, slug, keywords, category, meta), and produce an outline before any prose is written. Triggers on phrases like "ICP pre-writing for [topic]", "set up the ICPro brief for [post]", "outline this ICPro post". Output: `blog-posts-pre-writing/YYYY-MM-DD_[slug]_pre-writing.md` in the icpro-blog channel folder.

When the user wants the pre-event blog post that introduces a webinar topic and speakers. Trigger phrases include "webinar blog", "pre-event blog", "blog intro", "webinar blog post", "intro blog for webinar", "write the webinar blog". Run this AFTER the brief is filled out.

When the user wants webinar LinkedIn promo posts — intro post, partner mirror, partner thought-leadership post, last-minute boost posts. Trigger phrases include "linkedin posts", "linkedin promo", "linkedin webinar posts", "thought leadership post", "boost posts", "webinar social posts". Run this AFTER the brief is filled out.

When the user wants to identify short vertical clips (YouTube Shorts) from a webinar recording, plus LinkedIn cross-post copy — timestamps, captions, YouTube descriptions, and LinkedIn post copy for the video editor and social poster. Trigger phrases include "recap clips", "webinar shorts", "youtube shorts", "linkedin clips", "webinar clips", "key insights clips", "highlight clips", "post-webinar clips", "clip the webinar". Run this AFTER the cleaned transcript is available (i.e., after `webinar-ingest`).

When the user wants to draft, write, refine, or fill out a webinar brief. Trigger phrases include "write the brief", "draft the brief", "fill out the brief", "webinar brief content", "structure the outline", "polish the brief", "objective and core message". Run this AFTER `webinar-project-init` has scaffolded the brief skeleton AND the user has had at least one topic-defining conversation with the partner.

When the user wants registration page copy for a webinar — headline, subhead, value props, agenda, speaker bios, CTA. Trigger phrases include "registration page copy", "landing page copy", "webinar registration page", "reg page", "write the registration page", "polish the speaker bio". Run this AFTER the brief is filled out.

When the user wants a rehearsal checklist or day-of run-of-show / timing script for a webinar. Trigger phrases include "run of show", "runofshow", "rehearsal checklist", "rehearsal prep", "day-of script", "webinar timing", "live event prep", "webinar checklist". Run this AFTER the brief outline is settled, ideally a few days before the rehearsal.

Assign an owner to marketing contacts that have no owner. Ensures every marketable contact has accountability for follow-up, proper lead routing, and accurate owner-based reporting.

Enrich missing geographic data (country, state, city) on contacts and companies using HubSpot workflows, external data providers, or IP-based geolocation.

Build a workflow to protect sender reputation through automated bounce monitoring. Auto-suppresses contacts above a configurable bounce threshold, alerts on hard bounces, and flags high-bounce contacts for weekly manual review.

Create a comprehensive lead scoring model with separate Fit and Engagement scores using HubSpot's new Lead Scoring tool. Replaces the deprecated HubSpot Score property.

Create foundational segmented lists for marketing and sales operations. Includes a master sendable list, ICP-based lists, persona lists, engagement lists, and behavioral lists. All active (dynamic) lists.

Audit and consolidate HubSpot reporting dashboards. Identifies unused, duplicate, or outdated dashboards. Must be performed manually — no dashboard API is available.

Standardize deal pipelines, remove test deals, and address deals with missing amounts or close dates. Coordinates with Salesforce sync if applicable.