webinar-project-init

name	webinar-project-init
description	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.
metadata	{"version":"0.1.0"}

Webinar Project Init

Scaffold a new webinar event folder with the brief, project plan, and tracking-URLs files. Run this after the webinar topic and date have been agreed with the partner — it's the first thing Claude does once those two are locked.

Inputs to ask for

If any of these are missing, ask before proceeding. Don't guess.

1. Partner / speaker name — used to slug the folder (e.g., "Matt Frost" → matt-frost)
2. Event month and year — used in the folder name (<speaker-slug>-<month>-<YYYY>)
3. Event date — YYYY-MM-DD format
4. Event time — including timezone (e.g., "10:00 AM EST")
5. Webinar topic / working title (if known — placeholder OK)
6. Cerkl presenter (default to Tarek Kamil unless told otherwise)
7. External partner? Y/N — drives MAP customization in Step 4a. Y = standard partner co-marketing event; N = internal/solo (Cerkl-only speaker, no partner side). Infer from input #1 + #6 (e.g., speaker is Cerkl staff → N) but confirm if ambiguous.

Conventions to enforce

• Folder name: <speaker-slug>-<month-name-lowercase>-<YYYY>/ — e.g., matt-frost-april-2026/
• All dates inside files: YYYY-MM-DD (per cerkl/CLAUDE.md universal convention)
• File naming inside the folder: <speaker-slug>-<artifact>.md — matches the existing pattern in matt-frost-april-2026/

Speaker slug normalization

• Lowercase
• Strip titles and suffixes (Dr., PhD, Jr., etc.)
• Replace spaces with hyphens
• Drop punctuation (apostrophes, commas, periods)
• ASCII-only — transliterate accented characters (é → e, ñ → n)

Examples:

• "Matt Frost" → matt-frost
• "John O'Brien" → john-obrien
• "Dr. Jane Smith, PhD" → jane-smith
• "Crescenzo Communications" → crescenzo-communications
• "François García" → francois-garcia

Rehearsal date computation

Default rehearsal = Monday of the event week. Exception: if the event is on Monday, schedule rehearsal for Friday of the prior week. Override the T-3d offset in the project plan with this computed date.

Tasks to mark [x] (already done) when scaffolding

If the user is running this skill, the following tasks are by definition complete and should be pre-marked done in the project plan:

• "Finalize webinar topic and objectives" (a working title is in the brief)
• Any other task implied by the inputs the user provided

Don't pre-mark tasks that might be done — only the ones unambiguously implied by running webinar-project-init.

What to create

In cerkl/marketing/channels/webinar/<speaker-slug>-<month>-<YYYY>/:

1. <speaker-slug>.md — the brief

Copy from templates/brief-template.md. Pre-fill the metadata header:

• Title (from input, or leave placeholder)
• Date (YYYY-MM-DD)
• Time
• Featuring ([Cerkl presenter] and [Partner name + role]; for solo/internal, just the speaker)

Then substitute the "What we need from you" asks block tokens — the block is pre-populated in the template but with [BRACKET_TOKENS] that init must resolve so the Doc is send-ready as soon as it lands in Drive. Substitutions:

Token	Value
[SPEAKER_FIRST_NAME]	First name of input #1 (speaker) — used in the greeting
[EVENT_DATE]	Input #3, human-friendly format (e.g., June 25 or 2026-06-25) — used in both heading and Q1
[EVENT_TIME]	Input #4 with timezone (e.g., 12:00 PM EDT) — used in Q1
[CONTENT_DEV_DATE]	Event date − 21 days (the T-3w content-dev lock date) — used in the working-deadline line
[WORKING_TITLE]	Input #5 (working title) or TBD if not yet set — used in Q3

Leave all other placeholders (in the structured brief sections below the asks block) intact for the webinar-brief skill to populate later from the speaker's answers.

2. <speaker-slug>-project-plan.md — the task list

Copy from templates/project-plan-template.md. Then:

• Fill in Webinar Date (YYYY-MM-DD)
• Fill in Brief reference (<speaker-slug>.md)
• Fill in Collaboration Partner name
• Convert all T-Nw / T-Nd / T+N offsets to absolute YYYY-MM-DD dates based on the event date
• Update the <speaker-slug> placeholders in skill references and notes

3. <speaker-slug>-tracking-urls.md — Zoom URL inventory

Copy from templates/tracking-urls-template.md. Then:

• Fill in Webinar title and Webinar Date
• Replace <partner-slug> with the actual partner slug throughout the partner-side table (internal/solo: drop the partner-side table or note N/A)

URL slots stay empty — the user fills them in after creating each one in the Zoom portal.

4. Drive artifacts — MAP + Google Doc brief

Stand up two Drive documents and write their URLs back into the local .md files. Every init must produce both.

4a. Copy & customize the Mutual Action Plan

Source template (read-only — do not edit): 1cO4KvGdfcQIvnJxQ5orMLoh0GzQ0d78zl2C2nTzhY_c ("Cerkl Webinar Mutual Action Plan - Template"). Has one Project Plan tab matching the markdown plan structure (Category, Task, Priority, Owner, Status, Due Date, Deliverable, Notes).

1. Copy the template via Drive API:

   gws drive files copy --params '{"fileId":"1cO4KvGdfcQIvnJxQ5orMLoh0GzQ0d78zl2C2nTzhY_c"}' --json '{"name":"<NAME>"}'
   

   Naming, by partner branch (input #7):

   • Y (external partner): Cerkl Webinar Mutual Action Plan — <Speaker>, <Month> <YYYY>
   • N (internal/solo): Cerkl Webinar Internal MAP — <Speaker>, <Month> <YYYY>

   Capture the returned id. URL: https://docs.google.com/spreadsheets/d/<id>/edit

2. Clear the Project Plan tab:

   gws sheets spreadsheets values clear --params "{\"spreadsheetId\":\"<id>\",\"range\":\"'Project Plan'!A1:H50\"}"
   

3. Rewrite the Project Plan tab with gws sheets spreadsheets values update (range 'Project Plan'!A1, valueInputOption: RAW). Use the column structure above and the phase structure of templates/project-plan-template.md. Apply:

   • Header rows: Webinar Date → event date (e.g., Thu, June 25, 2026); Collaboration Partner → partner name (Y) or None — internal solo webinar (<Speaker>, Cerkl) (N).
   • Due Date column: convert all T-Nw / T-Nd / T+N offsets to absolute YYYY-MM-DD based on the event date (same conversion you applied to <speaker-slug>-project-plan.md).
   • Status column: pre-mark Complete for "Finalize webinar topic and objectives" and "Confirm date, time and platform" (both are by-definition done once init runs).
   • Internal/solo branch (N): remove all partner-owned rows (partner Email Blasts, partner LinkedIn posts, partner reminder rows, partner-side blog row); reassign mixed Cerkl/partner rows to Cerkl (bio + headshot, outline collaboration, downloadable asset); remove "Agree on speaker compensation".

   Worked example: the 2026-06-01 update on personal-assistant/projects/webinars.md (Rachel Folz webinar) — the JSON body written to values update is the structural reference for both branches; the solo branch is what was actually run.

4b. Upload <speaker-slug>.md as a Google Doc

Follow md-to-drive. One file, so run inline (no subagent). Override the default naming because the brief's H1 is the generic "Webinar Brief":

• Source: <event-folder>/<speaker-slug>.md
• Destination: default (Claude-Uploads, folder ID 1L4GEISXbi9sbqOKDCwbFuD1F6MF-JI5g)
• Name override: Cerkl Webinar Brief — <Speaker>, <Month> <YYYY>
• Cleanup: none

Capture the returned Doc URL: https://docs.google.com/document/d/<id>/edit.

4c. Write URLs back into the local .md files

Both local files have placeholder lines populated from the templates. Replace:

• In <speaker-slug>.md header:
  • **Mutual Action Plan:** [Drive link — filled in by ...] → [<Doc/Sheet display name>](<MAP URL>)
  • **Google Doc Brief:** [Drive link — filled in by ...] → [<Doc display name>](<Doc URL>)
• In <speaker-slug>-project-plan.md header: same two lines.

5. Canva asset manifests — <event-folder>/canva-manifests/

Stand up one template-fill manifest per role in the chosen pack from ../../asset-packs.md. The manifests are render-ready skeletons — they pre-fill what's known at init (title, date, time, speaker name, speaker role) and leave the headshot slot as TBD. They are not invoked at init; render happens later (Phase 2/3) once the speaker has provided a headshot.

5a. Pack selection

• Default: IC Thought Leadership pack (9 templates)
• Override: switch to Generic pack (5 templates) only if the user signals non-series styling. Confirm at scaffold if ambiguous.

5b. Generate manifests

Create <event-folder>/canva-manifests/ and write one YAML per role. Two-step process per role:

Step 1 — common skeleton. Start with this template:

template_id: <from pack table>
page_index: 1                            # override if pack notes otherwise
design_title: <speaker-slug>-<role>-<event-date YYYY-MM-DD>
text_values:
  "Webinar title": "<input #5 — working title>"
  "Name": ["<input #1 — speaker name>"]
  "Title": ["<input #6 — speaker role; e.g. 'Head of Product, Cerkl'>"]
  "|   Mar 6, 2025 1:00 pm est": "|   <Mon DD>, <YYYY> <h:mm AM/PM ZONE>"
image_assets:
  headshot: ["TBD — gather from speaker; resolve via canva-asset-index"]
commit_mode: checkpoint

Use lowercase "am/pm" if matching the template's casing convention; uppercase zone (e.g., EDT, EST).

Step 2 — apply per-role overrides from ../../asset-packs.md "Per-role manifest overrides" section. Each role's override block lists set / drop operations against text_values and image_assets. Apply them in order before writing the manifest to disk. Examples of why this matters:

• countdown → override locks "X": "2" (countdown is always the LinkedIn 2-days-to-go boost — no variability), drops vestigial Name/Title/headshot keys, and switches the date placeholder to the no-pipe variant the countdown template actually uses
• recap-blog-cover and recording-thumbnail → drop the date placeholder (those templates have no date element on page 1)
• Roles with no override block in asset-packs.md use the common skeleton as-is.

The exact placeholder per template lives in template-fill/_element-maps/<template_id>.json. Minor variance is handled by the orchestrator's smart matching at render time — see webinar-canva-render Step 4 (live richtexts[] exact-match override of template-fill's blind substring fallback).

5c. Write a canva-manifests/README.md index

List each role → manifest file → template ID → aspect ratio → promo step it serves (mirror the table in asset-packs.md). Include a one-line "to render" instruction.

5d. Render trigger (not run at init)

The manifests are dormant until the speaker provides the headshot. When ready, invoke template-fill per manifest. Each returned Canva edit URL gets pasted into:

• The matching row of the Drive MAP (Phase 3 promo row)
• The matching row of <speaker-slug>-project-plan.md (Phase 3, asset-role column)

After scaffolding — surface these reminders to the user

Print a checklist of next manual steps the user must take outside of Claude:

1. Zoom webinar — create in the Zoom portal. Then create one tracking registration URL per slug in <speaker-slug>-tracking-urls.md and paste them into the file. Reference: tracking-urls-convention.md.
2. Headshot upload — get the speaker's headshot uploaded as a Canva asset; capture the asset ID. Until this is done, the canva-manifests/ are dormant and template-fill won't render cleanly.
3. Canva render — once headshot is in hand, invoke template-fill per manifest in canva-manifests/. Paste returned URLs into the Drive MAP and project-plan rows.

Then list the next Claude skills to invoke as the project progresses (do not invoke them now):

• webinar-brief — fill out the brief content with the partner
• webinar-registration-page — once the brief is ready
• webinar-promo-emails — once the registration page is live
• webinar-linkedin-posts — paired with promo emails
• webinar-blog-intro — pre-event blog
• webinar-runofshow — before rehearsal
• webinar-ingest — immediately after the live event, once raw transcript / chat / deck are saved into <event-folder>/raw/. Produces the cleaned source-of-truth files every other post-event skill consumes.
• webinar-followup-email — day after the event (reads from the cleaned transcript)
• webinar-recap-clips — within a week of the event (reads from the cleaned transcript)
• webinar-recap-blog — within ~1–2 weeks; standalone topic-led blog post for the SEO archive (reads from the cleaned transcript + deck extract)

Push update

After scaffolding, append an update block to the relevant file in personal-assistant/projects/ per the protocol in ../../CLAUDE.md.