Skip to content
Baseworks KB

Asia's Inbox — Archive

Created 2026-03-05
Updated 2026-04-10
Tags inboxasiaarchive

Completed inbox items with full original content. Newest at the top.

Index

Section titled “Index”
2026-04-10
2026-04-08
2026-04-07
2026-04-06
2026-04-05
2026-04-01
2026-03-25
2026-03-19
2026-03-18
2026-03-14
2026-03-12
2026-03-09
2026-03-08
2026-03-07
2026-03-05
2026-03-04

2026-04-09 10:30 ET — New ACF field needed: “Last Updated” date on article post type

Section titled “2026-04-09 10:30 ET — New ACF field needed: “Last Updated” date on article post type”

Added by Claude Code on Patrick’s Mac

What: Articles on baseworks.com currently show only the publish date. When an article receives a significant revision (like the March 2026 proprioceptive awareness update), there’s no way to display that on the page. We need a new custom field and a template update.

What to create:

  1. New ACF field: article_last_updated — Date Picker type, attached to the article post type
    • Leave blank for articles that haven’t been significantly revised (only the original publish date shows)
    • Set manually only when a substantive revision is published (new data, revised argument, added/removed sections) — not for minor typo or formatting fixes
  2. Template display: Update the Elementor article template to show the field when set, formatted as: “Originally published [post date] · Updated [last_updated date]”
  3. First use: Set it on the proprioceptive awareness article to March 2026

Why manual rather than using WordPress post_modified: The built-in modified date updates on every save, including trivial edits. A manual field gives editorial control over what counts as a meaningful update — which is exactly how blog archives should work. Only significant revisions get timestamped; minor fixes don’t mislead readers.

Standard practice: Blog archives show “Published” + “Last updated” (two dates). No need for a full revision log — that lives in the vault and git history.

After creating the field: Let Patrick know so the /publish-article skill workflow can be updated to include this field.

Completed: 2026-04-10

2026-04-07 17:30 ET — Forum content sync live + moderation plan — your review needed

Section titled “2026-04-07 17:30 ET — Forum content sync live + moderation plan — your review needed”

Added by Claude Code on Patrick’s Mac

Two things here: a new system that’s already running, and a plan that needs your input.

1. Forum content ingestion is live (System 3)

Section titled “1. Forum content ingestion is live (System 3)”

All forum content from practice.baseworks.com now syncs into the vault automatically every 2 hours. This is the same architecture as the questionnaire export (System 1) — Python script on the VPS pulling from BuddyBoss REST API, generating markdown, committing to git.

What’s in the vault now:

02-areas/practice-platform/community-forums-groups/
├── community-posts/          ← pre-existing manual posts (unchanged)
├── forums/                   ← 4 forum entities (one file per forum + index)
│   ├── community.md
│   ├── hang-out.md
│   ├── montreal-study-group-2026-spring-cohort-forum.md
│   ├── montreal-study-group-2026-winter-cohort.md
│   └── practice-community-discussion.md
├── topics/                   ← 32 discussion topics (replies inline in each file)
│   ├── 250805-forum-guidelines.md
│   ├── 260124-forum-tag-shortlist.md
│   ├── 260124-session-1-summary-...md
│   ├── ... (all session summaries, questions, discussions)
│   ├── 260327-recu-facture-preuve-dachat.md   ← example: billing question
│   └── 260406-session-1-summary-foundations-first-contact.md
├── groups/                   ← 4 BuddyBoss groups with metadata + member counts
│   ├── montreal-study-group-winter-2026-cohort.md
│   ├── montreal-study-group-spring-2026-cohort.md
│   ├── practice-community.md
│   └── primer-community.md
└── activity/                 ← 5 monthly digests of group feed posts
    ├── 2021-01-activity-digest.md
    ├── ... through ...
    └── 2022-02-activity-digest.md

Key design decisions — please review these:

  • Replies are inline within topic files (under a ## Replies section), not separate files. This keeps each discussion as one readable document. If you think replies should be separate, let us know.
  • Topics are named {YYMMDD}-{slug}.md — date prefix for chronological sorting.
  • Activity digests are monthly — group feed posts aggregated by month, filtered for content-bearing items only (not logins or profile changes).
  • Groups get their own files — structural metadata (name, description, member count, forum ID) useful for wikilinks and context.

Each file has structured frontmatter with type, IDs, author, dates, reply counts, and source URLs back to the live forum. Claude can now search and cross-reference all forum content when working on session summaries, communications, or curriculum.

Documentation:

  • Full design: 03-resources/executed-plans/forum-content-ingestion-plan.md
  • System 3 in: 02-areas/website/functionalities/vault-sync-systems.md
  • Scripts: scripts/forum-content-sync.py and scripts/forum-content-sync.sh (also on VPS)

2. Forum moderation + content extraction plan — needs your review

Section titled “2. Forum moderation + content extraction plan — needs your review”

Patrick wants to build a moderation layer on top of this. The problem: participants occasionally post billing questions, receipt requests, consent concerns, or other admin matters to the public forum. Examples already in the vault:

  • Any Guay (2026-03-27): asked for her receipt/invoice publicly in the forum
  • Clementine (2026-03-14): posted a media consent concern on the group feed, which Patrick addressed with a public post redirecting admin matters to email

The plan has two parts:

Part 1 — Moderation (needs your input):

  • Detect off-topic posts using keyword matching (billing, payment, receipt, invoice, facture, consent, promotion, etc.)
  • Flag them for human review (notification to you and Patrick)
  • On approval: remove the post from the live forum + send a polite email to the poster explaining the forum is for course content only and their concern has been moved to email
  • Post preserved in the vault with status: moderated for record-keeping

Part 2 — Content extraction:

  • Analyze on-topic posts for programming-relevant insights (confusion points, breakthroughs, form-specific discussion, practice questions)
  • Generate periodic reports that feed into session prep and curriculum planning

Full plan: 03-resources/community/forum-moderation-plan.md

What we need from you

Section titled “What we need from you”
  1. Review the vault structure — browse the community-forums-groups/ directories in Obsidian. Does the organization make sense? Would you change anything?
  2. Review the moderation categories — are the five categories (billing, terms, consent, promotions, off-topic personal) complete? Anything missing?
  3. Email template tone — the plan includes a template approach for the removal notification emails. What address should these come from? Any tone notes?
  4. Content extraction signals — what insights from forum posts would be most useful for your programming decisions?
  5. Open question: should moderated posts stay in the vault (with a moderated flag) or be removed entirely?

Completed: 2026-04-08

2026-04-07 13:30 ET — Vault infrastructure upgrade — what changed and why

Section titled “2026-04-07 13:30 ET — Vault infrastructure upgrade — what changed and why”

Added by Claude Code on Patrick’s Mac

This is a summary of the infrastructure work done today so you understand what changed and why. No action needed from you — the setup item above covers your part.

Why we did this

Section titled “Why we did this”

As we add more content to the knowledge base (session summaries, campaigns, forum content, emails, curriculum materials), Claude needs to efficiently find and cross-reference information. We had two problems:

  1. No structural search. qmd can find content by meaning (“find documents about body awareness”), but it can’t answer structural questions like “what files link to the voice guide?” or “which files share the ‘curriculum’ tag?” These structural queries are important when building campaigns, writing session summaries, or checking how content connects across areas.

  2. Stale content between machines. The Obsidian Git plugin committed changes locally but didn’t push automatically. Changes made on one machine could take hours to reach others, causing drift.

What changed

Section titled “What changed”

1. CLAUDE.md is leaner now (225 → 175 lines)

We removed reference material (backup commands, machine identity tables, skill creation steps) and moved it to the files where it belongs. All the behavioral rules that Claude follows at session start are unchanged. This makes Claude more reliable — fewer instructions competing for attention means each one gets followed more consistently.

2. SQLite relationship index (vault-index.db)

A new database that indexes every wikilink, tag, and frontmatter field in the vault. It rebuilds automatically after every sync. Claude can now instantly query:

  • What files link to a specific file (backlinks)
  • What files share a tag
  • Which files are orphans (nothing links to them)
  • Which wikilinks are broken

This complements qmd — qmd finds content by meaning, the index finds content by structure.

3. Automated vault sync (every 5 minutes) — replaces Obsidian Git

This is the fix for the sync failures we kept having with Obsidian Git. A new script (scripts/vault-sync.sh) replaces the Obsidian Git auto-commit entirely. Here’s how it works in practice:

  • Every 5 minutes, each machine (your Macs, Patrick’s Mac, and the VPS) automatically pulls changes from GitHub, commits any local changes, and pushes. You do not need to do anything — no manual commits, no remembering to push.
  • If your Mac is asleep, the sync pauses. The moment it wakes up, it fires immediately — pulls everything that was pushed while you were away, commits and pushes anything you had locally. No changes are lost, just slightly delayed.
  • The VPS never sleeps. It runs 24/7, so automated systems (forum content sync, questionnaire export) always push on schedule regardless of whether anyone’s Mac is on.
  • During Claude Code sessions, every file edit is written to disk instantly. The sync picks it up within 5 minutes. Even if a terminal session crashes mid-work (as happened today), whatever was written to disk before the crash is safe — the next sync cycle commits and pushes it.
  • Why Obsidian Git failed: it committed locally but didn’t push reliably. Changes would sit on one machine for hours. The new system does the full pull-commit-push loop every cycle, so drift between machines is eliminated.

You’ll see commits in the git history labeled vault sync: (new system) alongside older vault backup: commits (Obsidian Git). Both are fine. Once we confirm everything is stable across all machines, we’ll disable Obsidian Git completely.

Changes now propagate between machines within 5-10 minutes instead of waiting for manual pushes. After each pull, the system also automatically rebuilds the SQLite index and updates qmd.

4. /vault-audit health check

A new skill that checks the health of the entire system — CLAUDE.md size, index freshness, qmd status, vault structure, sync status, and growth trends. Run it anytime with /vault-audit to see if anything needs attention.

What this means for your workflow

Section titled “What this means for your workflow”

Nothing changes about how you use Obsidian. You still edit files normally. The sync happens in the background. The only visible difference is that commits will now say vault sync: instead of vault backup: — both are fine, they’ll coexist until we fully transition.

What we evaluated and skipped

Section titled “What we evaluated and skipped”

We looked at LightRAG (a graph database system for AI retrieval) and related tools. They’re designed for thousands of documents with complex relationships. Our vault has ~754 files with explicit wikilinks — the SQLite index gives us the same relationship awareness at zero ongoing cost. We’ll revisit if the vault grows past ~2,000 files.

Completed: 2026-04-08

2026-04-07 11:05 ET — Vault cleanup — rogue files removed, folders reorganized

Section titled “2026-04-07 11:05 ET — Vault cleanup — rogue files removed, folders reorganized”

Added by Claude Code on Patrick’s Mac

Patrick audited the vault for out-of-place files and folders at the root level. Several empty stubs, orphaned files, and misplaced folders were cleaned up. Here’s what changed:

Deleted (empty/orphaned)

Section titled “Deleted (empty/orphaned)”
  • Style Guide.md — 0 bytes, empty placeholder
  • deployment-guide.md — 0 bytes, empty placeholder
  • index.md — 0 bytes, leftover from old Quartz migration
  • yogajaya-integration-awareness-transformation.webp — orphaned image, no note linked to it
  • patrick-oancia/ — folder with single empty stub file
  • study-groups/ — folder with single empty stub file (real content is in 02-areas/educational-programs/study-groups/)
  • assets/*.png — 5 untracked timestamp-named screenshots (previously cleaned up, had reappeared)

Reorganized

Section titled “Reorganized”
  • Downloads/ (winter-2026 closing presentation PDFs + exported slides) → moved to 04-archive/2026-winter-closing-presentation/

Obsidian config

Section titled “Obsidian config”
  • Added site/, _static/, .github/ to Obsidian’s excluded folders so they no longer pollute vault search or graph view. These are build infrastructure, not knowledge content.

No action needed from you

Section titled “No action needed from you”

This is informational — just so you know the vault is a bit tidier. The PARA structure (00-04 folders) and all content files are untouched.

Completed: 2026-04-08

2026-04-07 13:00 ET — Vault sync upgrade — one-time setup needed on your machines

Section titled “2026-04-07 13:00 ET — Vault sync upgrade — one-time setup needed on your machines”

Added by Claude Code on Patrick’s Mac

We’ve replaced the Obsidian Git auto-commit with a dedicated sync script (scripts/vault-sync.sh) that pulls, commits, and pushes every 5 minutes. This is more reliable — changes reach GitHub faster and you get other people’s changes pulled automatically.

A SQLite relationship index (vault-index.db) has also been added — it indexes all wikilinks, tags, and frontmatter so Claude can quickly query backlinks, tag membership, and orphan files.

The VPS is already configured. Your machines need a one-time setup:

Setup steps (per machine)

Section titled “Setup steps (per machine)”
  1. Open Terminal and run:
Terminal window
cd ~/Obsidian/baseworks-kb-shared-brain
git pull origin main
git config core.hooksPath scripts/git-hooks
python3 scripts/build-vault-index.py --full
cp scripts/launchd/com.baseworks.vault-sync.plist ~/Library/LaunchAgents/
launchctl load ~/Library/LaunchAgents/com.baseworks.vault-sync.plist
  1. Verify it’s running:
Terminal window
launchctl list | grep vault-sync
tail -5 ~/logs/vault-sync.log

Machines required:

  • Mac Mini
  • MacBook Air (M1 Laptop)

Important: Do NOT disable Obsidian Git yet — it will continue to work alongside the new sync until we confirm everything is stable on all machines. Once confirmed, we’ll disable the Obsidian Git auto-commit across all machines in a single coordinated push.

Requested action: Run the setup steps on each of your machines and check off the boxes above.

Completed: 2026-04-07

2026-04-05 23:50 ET — Staging sites causing 502s on production — Automator crons need attention

Section titled “2026-04-05 23:50 ET — Staging sites causing 502s on production — Automator crons need attention”

Added by Claude Code on Patrick’s MacBook Pro

Tonight Patrick was getting intermittent 502 Bad Gateway errors on practice.baseworks.com. After investigating, we traced the issue to the staging sites (staging.baseworks.com and pracstage.baseworks.com) consuming excessive CPU on the shared server, starving production of resources.

What we found

Section titled “What we found”

The staging site has three Uncanny Automator / Action Scheduler WP cron hooks that fire every 1–2 minutes:

HookInterval
automator_wp_post_published_in_taxonomy_posts_publishedEvery 1 minute
automator_userspost_posts_publishedEvery 1 minute
action_scheduler_run_queueEvery 1 minute

When the system cron triggered wp cron event run --due-now (every 5 minutes), these hooks would all fire, spawning PHP workers that each consumed 33%+ CPU. The staging PHP-FPM pool was configured with pm.max_children=20 (higher than production), so it could spawn up to 20 workers at once. This starved production sites of CPU, causing intermittent 502 errors.

What we fixed tonight

Section titled “What we fixed tonight”
  1. PHP-FPM pool limits reduced on both staging and pracstage:
    • Process manager changed from dynamic to ondemand (workers spawn on demand, die when idle)
    • max_children reduced to 3 (from 20 on staging, 6 on pracstage)
    • process.priority = 15 added so the OS always prioritizes production workers
    • memory_limit reduced to 256M (from 512M / 1024M)
  2. System cron frequency reduced from every 5 minutes to every 30 minutes for both staging sites

What needs your attention

Section titled “What needs your attention”

The Automator crons on staging.baseworks.com are still active — they just run less frequently now and with fewer resources. Since these are automations you manage, could you:

  1. Review whether Uncanny Automator needs to be active on staging at all. If the automations are only relevant to production, deactivating the Automator plugin on staging would eliminate the issue entirely.
  2. If Automator must stay active on staging, check whether the 1-minute recurrence intervals on those three hooks can be increased (e.g., to hourly or daily). These intervals are likely inherited from the production configuration when staging was cloned.

No urgency — the server-side limits we applied tonight will prevent staging from impacting production going forward. This is about cleaning up the staging configuration so it runs more efficiently.

Reference

Section titled “Reference”

Full details in the changelog: CHANGELOG.md entry for 2026-04-05.

Completed: 2026-04-06

2026-04-05 23:05 ET — Session 1 — edits accepted, forum post live

Section titled “2026-04-05 23:05 ET — Session 1 — edits accepted, forum post live”

Added by Claude Code on Patrick’s Mac

Patrick accepted all of your edits to the Session 1 summary. The post is now live on the group forum:

The group post (assignment announcement + summary link + Primer assignments reference) has also been finalized and is ready to copy-paste from the summary file:

No action needed — informational only.

Completed: 2026-04-06

2026-04-03 16:50 ET — Session 1 prep notes for tomorrow — please review

Section titled “2026-04-03 16:50 ET — Session 1 prep notes for tomorrow — please review”

Added by Claude Code on Patrick’s Mac

Patrick put together session preparation notes for tomorrow’s first Spring Study Group session (April 4, 9:10-10:50 AM, Studio 1). The notes are based on the full arc of the Winter study group session summaries — what worked, what to plant earlier this time, and a suggested time allocation.

Files to review:

Key points in the prep notes:

  1. Same three foundational forms as Winter Session 1 (Squat, Star Form, Star Tilt) — the Winter arc confirmed this is the right amount
  2. Suggestion to plant the moderation seed earlier than Winter (where it didn’t become explicit until Session 3)
  3. Brief forum orientation (~5 min) covering group feed vs. discussion forum distinction
  4. Name the 7-session arc upfront so participants know the trajectory
  5. Common corrections to watch for, drawn from Winter Session 1

Requested action: Review the session 1 prep note and flag anything you’d change or add before tomorrow morning.

Completed: 2026-04-05

2026-04-03 10:15 ET — Perception Gap blog — LinkedIn campaign published

Section titled “2026-04-03 10:15 ET — Perception Gap blog — LinkedIn campaign published”

Added by Claude Code on Patrick’s Mac

LinkedIn article, personal feed post, and Baseworks company page post are all published for the Perception Gap blog campaign. Campaign file updated with LinkedIn article URL, SEO metadata, and all three post texts. Hyperlinks added to the LinkedIn article body matching the baseworks.com version.

All LinkedIn social-status fields marked published in the campaign frontmatter. Instagram and Facebook remain at ready status.

[ ] 2026-04-02 17:10 ET | Welcome email HTML ready for FluentCRM automation — new Practice Session signups

Section titled “[ ] 2026-04-02 17:10 ET | Welcome email HTML ready for FluentCRM automation — new Practice Session signups”

Added by Claude Code on Patrick’s Mac Updated 2026-04-02 17:10 ET — final version with design refinements

The welcome email for new introductory Practice Session participants is finalized and ready to be set up as a FluentCRM automation.

HTML file: 02-areas/communications/email-templates/practice-sessions/welcome-new-assets/welcome-new-final.html

Copy the full HTML into a FluentCRM email action block. It is self-contained (inline styles, no external dependencies beyond Google Fonts and CDN images).

Automation trigger: Purchase of introductory session ($44 CAD) — tag track-introductory applied.

Subject line: Your introductory session — what to know before you arrive Preheader: Three preparation points, your calendar subscription, and the event participation questionnaire.

What the email contains:

  1. Hero image + welcome text + three preparation points (fresh eyes, follow instructions, natural breathing)
  2. CTA button: “Read the Preparation Guide” → practice.baseworks.com/preparation-guide/
  3. Calendar photo + styled Google Calendar and Apple/Outlook subscription buttons
  4. Questionnaire mock image + CTA button: “Complete the Questionnaire” → practice.baseworks.com/members/me/background/
  5. Closing line: “If you have questions before your session, reply to this email.”
  6. Standard footer with social links, unsubscribe, view in browser

CDN images used:

  • media.baseworks.com/website/practice-sessions/baseworks-what-to-expect.jpg — hero
  • media.baseworks.com/website/practice-sessions/baseworks-calendar-practice.jpg — calendar section
  • media.baseworks.com/website/practice-sessions/baseworks-event-questionnaire.jpg — questionnaire mock

FluentCRM merge tags already in the HTML:

  • ##crm.unsubscribe_url## — unsubscribe link
  • ##web_preview_url## — view in browser link

Full documentation: Welcome Email — New Participants

Completed: 2026-04-05

2026-03-31 21:00 ET — Set up B2 authorization for /compress-photos on your Mac Mini

Section titled “2026-03-31 21:00 ET — Set up B2 authorization for /compress-photos on your Mac Mini”

Added by Claude Code on Patrick’s Mac

The /compress-photos skill runs entirely on the local Mac — it processes images with sips, cwebp, and exiftool, then uploads directly to Backblaze B2. To use it from your machine, you need b2 CLI installed and authorized for the baseworks-media bucket.

Step 1: Install dependencies (if not already present)

Section titled “Step 1: Install dependencies (if not already present)”
Terminal window
brew install b2-tools webp exiftool

(sips is built into macOS — no install needed.)

Step 2: Create a new B2 application key for your Mac

Section titled “Step 2: Create a new B2 application key for your Mac”
  1. Log into Backblaze at secure.backblaze.com with the top-level admin account
  2. Go to Application Keys (left sidebar under “Account”)
  3. Click Add a New Application Key
  4. Settings:
    • Name: asia-mac-mini-baseworks-media
    • Allow access to Bucket(s): select baseworks-media only
    • Type of Access: Read and Write
    • Leave everything else default
  5. Click Create New Key
  6. Copy both the keyID and applicationKey immediately — the application key is only displayed once. Save them somewhere secure.

Step 3: Authorize b2 on your Mac

Section titled “Step 3: Authorize b2 on your Mac”
Terminal window
b2 account authorize <keyID> <applicationKey>

Step 4: Verify it works

Section titled “Step 4: Verify it works”
Terminal window
b2 ls b2://baseworks-media 2>/dev/null | head -5

You should see files listed. After that, /compress-photos will work from your Mac exactly as it does on Patrick’s.

Note on per-machine keys

Section titled “Note on per-machine keys”

Each machine gets its own application key scoped to just the baseworks-media bucket. Patrick’s Mac already has one. We will create a separate one for the VPS later. This way if any single key is compromised, we revoke just that key without affecting the others.

Do NOT reuse Patrick’s key or share keys between machines.

[ ] 2026-03-31 20:30 ET | KB site infrastructure needs sync — VPS site/ repo diverged + .quartz-dev has no remote

Section titled “[ ] 2026-03-31 20:30 ET | KB site infrastructure needs sync — VPS site/ repo diverged + .quartz-dev has no remote”

Added by Claude Code on Patrick’s Mac

Two issues discovered today that need attention:

1. VPS site/ repo (Astro + Starlight) is diverged with merge conflicts On the agents VPS (patrick@46.224.129.16), the repo at /srv/baseworks/knowledge-base/site/ has diverged: 2 local commits vs 174 remote commits, with unmerged paths. This means the VPS KB site build is significantly out of sync with the GitHub repo. The merge was started but hit conflicts and was never resolved.

Action needed: SSH to the VPS, assess the conflicts (git diff --name-only --diff-filter=U), determine if the 2 local commits contain changes worth preserving, and either resolve the merge or reset to remote. Be careful — this is the build that deploys kb.baseworks.com.

2. .quartz-dev on Patrick’s Mac has no Baseworks remote The .quartz-dev folder at the vault root is in detached HEAD state. The only remote is origin → jackyzha0/quartz.git (upstream open-source Quartz). There’s no Baseworks fork or remote configured. It has 8 uncommitted local changes: quartz.config.ts, quartz.layout.ts, custom.scss, path.ts, logos, and content symlink.

Questions:

  • Is .quartz-dev still needed alongside the Astro + Starlight setup?
  • If yes, should we create a Baseworks fork on GitHub and set it as the remote?
  • If no, should we remove the submodule from the vault?

Patrick is going to try to address the VPS issue in a new session now. Any context you have on the intended state of either setup would help.

[ ] 2026-03-31 16:15 ET | New /publish-article skill created + blog post publishing workflow documented

Section titled “[ ] 2026-03-31 16:15 ET | New /publish-article skill created + blog post publishing workflow documented”

Added by Claude Code on Patrick’s Mac

A new Claude Code skill /publish-article has been created for publishing finalized blog posts from the Obsidian vault to baseworks.com. This is the technical counterpart to /write-blog-post (which handles content creation in the vault) and /compress-photos (which handles media processing).

What it does: Takes a finalized blog post from the vault and publishes it to WordPress as an article post type. Handles uploading WebP images to the media library, converting markdown to Gutenberg blocks, setting ACF fields (hero image and subtitle), assigning categories and tags, configuring SEOPress meta, and verifying everything is correct. Always saves as draft for human review.

Skill file: .claude/skills/publish-article/SKILL.md Skills index: Claude Code Skills (table entry + detailed description added)

Key things documented in the skill (learned the hard way today):

  1. WP-CLI deprecation warning — the server outputs a PHP deprecation warning on every command. If you pipe wp post get --field=post_content back into wp post update without filtering it, the warning gets embedded in the post content and shows on the frontend. The skill documents how to filter it.
  2. Taxonomy assignment requires --by=id — without this flag, wp post term set 49080 category 365 creates a new term named “365” instead of assigning the Body Awareness category. Always use --by=id.
  3. ACF field keys — hero image is article_image (field key field_61f188ff7311c), subtitle is article_subtitle (field key field_68e6d4da57d3b). Both need the meta key AND the underscore-prefixed field key reference set.
  4. Post excerpt is separate from SEOPress description — both must be set.
  5. Category and tag ID lookup tables are included in the skill for quick reference.

Also done today:

  • Published “What You Can’t Feel, You Can’t Change: Body Awareness and the Perception Gap” (post ID 49080) — Patrick’s new article adapted from the March Brain Fodder
  • Added image border-radius CSS (10px all corners) to Kadence Customizer for all article content images (was previously only top corners)
  • Two campaign notes created: perception gap blog promotion + Spring Study Group launch at Proto Studio

Re: /compress-photos — if you’re working through the compress-photos workflow and have questions, the skill file is at .claude/skills/compress-photos/SKILL.md. The key steps are: (1) get images into NAS incoming, (2) run the skill with source folder and B2 destination, (3) it handles resize, WebP+JPEG compression, EXIF stripping, B2 upload, and NAS copy. Happy to walk through it in a session if helpful.

[ ] 2026-03-30 22:00 ET | Spring 2026 cohort communications ready — forum post, group feed post, and 7 emails

Section titled “[ ] 2026-03-30 22:00 ET | Spring 2026 cohort communications ready — forum post, group feed post, and 7 emails”

Added by Claude Code on Patrick’s Mac

All pre-course and between-session communications for the Spring 2026 cohort have been drafted and are ready in: Cohort Emails — Spring 2026

Action needed:

  1. Post the Primer Assignments to the Spring cohort forum — the full text is in the file above under “Forum Post: Primer Assignments.” Copy-paste into the forum, set as Sticky and Closed. This is the same format you posted for the Winter cohort on January 6.

  2. Post the group feed announcement — short post linking to the forum post. Title: “Primer Assignment Schedule — Spring 2026.” Text is in the file under “Group Feed Post.”

  3. Review Email 01 (Welcome and First Assignment) — Patrick wants to send this tonight or as soon as the forum post is live. The [here] link in the email needs the actual forum post URL once you’ve posted it.

  4. Emails 02-07 are session-by-session reminders, sent 3 days before each session (Wednesdays). All drafted and ready. These can be sent manually or set up as scheduled sends.

Other items from this session:

  • New contact inquiry from James Murray (Montreal, self-employed) about the Spring Cohort — response drafted and approved by Patrick. See James Murray inquiry.
  • Voice guide updated (v2.3 draft): “Thank you for reaching out” added to Avoid list — use “Thank you for your inquiry” instead. Pending your confirmation.
  • Contact inquiry response formatting changed: outbound responses now use plain text (not blockquotes) for easy copy-paste into email. Applied across all three inquiry files (Millie, Sylvia, James).

[x] 2026-03-29 02:30 ET | Any Guay CE inquiry — draft email response + participation letter ready for your review

Section titled “[x] 2026-03-29 02:30 ET | Any Guay CE inquiry — draft email response + participation letter ready for your review”

Added by Claude Code on Patrick’s Mac

Patrick has drafted two documents in response to Any’s latest emails (March 21 gratitude message, March 24 invoice request, March 27 association requirements). Both are ready for your review before sending.

Full context: Any Guay — Continuing Education Certificate Inquiry — the complete thread with all communications (emails, forum posts, group feed posts) documented in both French and English.

1. Email response to Any (EN + FR):

Acknowledges her March 21 gratitude message, reaffirms our position on CE documentation, offers to confirm hours in writing with a disclaimer, notes future credentialing-specific programs would be clearly designated. See “Draft Response (2026-03-28)” section in the linked document.

2. Participation letter for her association (“To Whom It May Concern”):

A factual record of her participation confirming sessions attended (5/7, ~9 of 12 hours) and online segments completed (7/10, ~8 of 12 hours). The letter explicitly states that:

  • The program was designed for personal enrichment and practical, non-professional application
  • Baseworks does not operate as a continuing education provider
  • This is an intentional organizational position, not an administrative limitation
  • The letter is provided as a factual record of participation only

Full text of the letter:

To Whom It May Concern,

This letter confirms the participation of Any Guay in the Baseworks Winter 2026 Study Group, held in Montreal, Quebec between January and March 2026.

Participation details:

  • In-person sessions attended: 5 of 7 (approximately 9 of 12 scheduled hours)
  • Online coursework completed: 7 of 10 segments of the Baseworks Primer (approximately 8 of 12 available hours)

About the program:

The Winter 2026 Study Group was an introductory program designed to familiarize participants with the core principles and practices of the Baseworks Method. The program combined weekly in-person group sessions with self-directed online coursework through the Baseworks Practice Platform.

This program was designed for personal enrichment and practical, non-professional application. It was not structured, accredited, or intended as a continuing education or professional development program within any credentialing framework.

About Baseworks:

Baseworks does not operate as a continuing education provider and does not issue continuing education credentials, certificates, or attestations. This is an intentional organizational position, not an administrative limitation. Our programs are designed around direct engagement with the material for its own value, independent of external credentialing systems. As such, no program offered by Baseworks should be interpreted as qualifying for professional continuing education credit unless explicitly stated otherwise in the program description.

This letter is provided as a factual record of participation only.

Patrick Oancia Founder, Baseworks baseworks.com

Requested action: Review both drafts. Send as-is or modify. The email response should go to Any; the letter would be attached or sent separately for her to forward to Trager Québec. Note: her association’s deadline is April 1.

Context on the March 24 email gap: Any emailed on March 24 asking for her invoice. This went unanswered, which is why she posted on the forum and group feed on March 27. Patrick only saw the March 27 association requirements email, not the earlier ones. This has been noted in the thread — no action needed beyond awareness.

Completed: 2026-04-01

2026-03-25 — Writing Process + Voice Guide Structural Changes — informational

Section titled “2026-03-25 — Writing Process + Voice Guide Structural Changes — informational”

Added by Claude Code on Patrick’s Mac

Patrick reviewed your feedback on the Winter 2026 blog post and approved most of it with voice corrections. This item was informational — your role and workflow did not change. Archived 2026-03-31.

What changed:

  1. AI Generation Warning added to top of VOICE-GUIDE-UNIFIED.md — explicitly names the failure mode (contrastive/deficit framing) and provides three filters every sentence must pass.

  2. WRITING-PROCESS.md created at 03-resources/voice-guides/WRITING-PROCESS.md — the shared process layer. Covers voice loading order, the three filters, feedback-to-proposal workflow, and semantic structure for AI discoverability.

  3. /write-blog-post and /create-campaign skills updated — both now load WRITING-PROCESS.md as step 1.

Voice loading order:

  1. WRITING-PROCESS.md — the shared process
  2. VOICE-GUIDE-UNIFIED.md — the shared voice (including AI Generation Warning)
  3. VOICE-GUIDE-KSENIA-ASIA.md — your personal voice
  4. Content-type guidelines

Files changed:

  • 03-resources/voice-guides/WRITING-PROCESS.md (new)
  • 03-resources/voice-guides/VOICE-GUIDE-UNIFIED.md (updated — AI Generation Warning, v1.9 confirmed)
  • .claude/skills/write-blog-post/SKILL.md (updated)
  • .claude/skills/write-page-copy/SKILL.md (new)
  • .claude/skills/create-campaign/SKILL.md (updated)
  • 03-resources/claude-code-skills/index.md (updated — write-page-copy added)

2026-03-25 — Voice guide v2.0 — feedback patterns — informational, no action required

Section titled “2026-03-25 — Voice guide v2.0 — feedback patterns — informational, no action required”

Added by Claude Code on Patrick’s Mac

Patrick reviewed your feedback on the Winter 2026 blog post and added several patterns to the unified voice guide (v2.0 draft). Informational — no action required. Archived 2026-03-31.

Patterns added:

  1. “Rather than” / negative framing — Drop “rather than” clauses. The positive statement is strong enough on its own. Voice guide principle 2: describe what something IS.

  2. Generated metaphors — “Running on autopilot” approved for occasional use (max once per piece). “Clicks into place” not approved — replaced with “becomes physically tangible.”

  3. “Transform” usage — Can be used occasionally in formal/deep copy, but default to “changes,” “shifts,” or “recalibrates.” Remains on Avoid list as general rule.

  4. Leading with the positive — “The idea is not repetition but re-encounter” → “The idea is re-encounter.” Lead with what something IS; clarify distinction in a separate sentence if needed.

  5. Smart Revisit and PrimerPrint — Always capitalize as named platform features.

All changes in VOICE-GUIDE-UNIFIED.md v2.0 draft, Good vs. Needs Revision table, and Allowed Metaphors section.

2026-03-18 16:45 ET — How to create and use Claude Code skills — informational, no action needed

Section titled “2026-03-18 16:45 ET — How to create and use Claude Code skills — informational, no action needed”

Added by Claude Code on Patrick’s Mac

A guide was added to the vault explaining how Claude Code skills work and how to create new ones. Informational — no action needed. Archived 2026-03-31.

Key files:

  • 03-resources/claude-code-skills/how-to-create-skills.md — full guide
  • 03-resources/claude-code-skills/index.md — lists all available skills

Short version: For any repeatable multi-step workflow, tell Claude: “I want a skill that does X.” Claude creates the skill file, guidelines document, and adds it to the index. After a commit and push, the skill is available on all machines via /skill-name.

2026-03-19 02:00 ET — baseworks-changelog repo — branch divergence, needs cleanup session

Section titled “2026-03-19 02:00 ET — baseworks-changelog repo — branch divergence, needs cleanup session”

Resolved 2026-03-19 11:00 ET — merge completed by Claude Code on Patrick’s Mac (see item above)

Summary: The baseworks-changelog GitHub repo has two branches (master and main) that have been diverging independently for months. Your Mac has been committing to main correctly — but the VPS has been committing to master in parallel. A cleanup session is needed to merge them into one branch.

Completed: 2026-03-19

2026-03-18 21:30 ET — Closing presentation posted to practice site — email sent to Winter cohort

Section titled “2026-03-18 21:30 ET — Closing presentation posted to practice site — email sent to Winter cohort”

Added by Claude Code on Patrick’s Mac

The closing session slide deck from March 1 has been posted to the practice site as a WordPress post with individual slide images and bilingual (EN/FR) summaries for each slide. The post also includes reminders about practice sessions (with booking links) and Primer timeline/expiry.

Post URL: https://practice.baseworks.com/info/cohort-updates/winter-2026-closing-presentation/ Category: Cohort Updates (new category created on the practice site)

An email was sent to all Winter 2026 cohort participants notifying them the presentation is available and inviting feedback via the forum or email.

Vault records:

No action required — this is informational. The post is live and the email has been sent. Review the post if you’d like to confirm the French text.

Completed: 2026-03-18

2026-03-18 19:00 ET — URGENT — Create individual taxonomy concept notes for the entire Baseworks taxonomy

Section titled “2026-03-18 19:00 ET — URGENT — Create individual taxonomy concept notes for the entire Baseworks taxonomy”

Added by Claude Code on Patrick’s Mac

Priority: Do this today.

What’s needed: Every Baseworks taxonomy concept needs its own standalone note in the vault — not just a section inside a large file. Right now, concepts like Distributed Activation, Micro-Movements, Star Form, etc. exist only as sections within 02-areas/method-admin/core/key-definitions.md, 02-areas/practice-platform/glossary-new.md, and 02-areas/method-admin/core/taxonomy.md. This means wikilinks like [Distributed Activation](/areas/method-admin/core/key-definitions-repo/distributed-activation/) are broken vault-wide because there’s no file with that name.

Why this matters:

  • Every campaign note, session summary, and Primer transcript that references a concept has a broken wikilink right now
  • The graph can’t show connections between concepts and the content that references them
  • QMD semantic search can’t retrieve individual concepts — they’re buried inside monolithic files
  • The new campaign infrastructure we built today depends on rich concept-level wikilinks to work as designed

What needs to be created (approximately 50-80 notes):

  1. Principles — Distributed Activation, Micro-Movements, Gridlines and Symmetry, Fixing-Separating-Isolating, Intensity Modification, Natural Breathing
  2. Forms — all 27+ forms from the taxonomy (Squat, Star Form, Star Tilt, Standing Form, etc.)
  3. Foci — Structure, Gravity, Ascend, Torsion, Converge, Expand, Inflect, Intent, Equate, Transit, Transpose, Isolate
  4. Framework concepts — Sense-Control-Adapt, Ignition, Assimilation, Cyclical Practice, Sensory Resolution, Physical Intelligence, Perceptual Skills, etc.

Source material to draw from:

  • 02-areas/method-admin/core/key-definitions.md — foundational definitions
  • 02-areas/practice-platform/glossary-new.md — current practice platform glossary
  • 02-areas/method-admin/core/taxonomy.md — standardized terminology for forms, foci, patterns
  • 02-areas/primer/transcripts-en/ — Primer concept transcripts (rich, polished explanations)
  • 02-areas/method-admin/core/philosophy/ — philosophy docs (Assimilation, Movement and Emotional Processing)
  • 02-areas/method-admin/core/science-docs/ — science docs (Iterative Refinement)

Patrick says you have a way you want to organize this. Please use your preferred structure. The key requirements are:

  • Each concept is its own .md file with a unique filename that matches how it’s referenced in wikilinks (e.g., Distributed Activation.md so [Distributed Activation](/areas/method-admin/core/key-definitions-repo/distributed-activation/) resolves)
  • Each note has proper frontmatter (title, created, tags)
  • Each note links back to related concepts, programs, and source material
  • Run python3 scripts/check-wikilinks.py --changed after creating to verify links resolve

Requested action: Start a Claude Code session and work through this today. You can organize the notes however you think is best — the only hard requirement is that the filenames match the wikilink targets so all existing broken links resolve.

Completed: 2026-03-18

2026-03-13 17:00 ET — Practice site infrastructure documented — review and discuss refactoring plan

Section titled “2026-03-13 17:00 ET — Practice site infrastructure documented — review and discuss refactoring plan”

Added by Claude Code on Asia’s Mac Mini

New note created at 02-areas/website/practice-site-platform-infrastructure.md documenting the full practice site automation infrastructure — dashboard sections, Formidable forms, activity tracking recipes, code snippets, and WPML setup.

Also includes a scalability audit and refactoring proposal for moving Form 69 (activity log) from Formidable’s EAV storage to a purpose-built custom database table. Short version: Formidable Forms will not scale well as an activity log past ~1,000 subscribers, especially once PrimerPrint is implemented online. A custom table + lightweight plugin + REST API is the recommended path.

Key questions to discuss:

  1. Do you agree with the refactoring direction?
  2. Timeline priority — should this happen before or after the Live Sessions feature is built?
  3. PrimerPrint online: do you want to plan this as part of the same plugin refactor, or as a separate initiative?

Also noted: Code snippets IDs 30–34 and 40 are not yet exported to the changelog repo. This should be done soon.

Completed: 2026-03-14

2026-03-11 11:40 ET — Claude Code VPS authentication — resolved

Section titled “2026-03-11 11:40 ET — Claude Code VPS authentication — resolved”

Added by Claude Code on Patrick’s Mac

Resolved: A long-lived token (valid 1 year) has been set up on the VPS for both patrick and asia users. No action needed from you — Claude Code should work when you SSH in. If you hit a 401 error, see 00-inbox/claude-vps-credential-sync.md for renewal steps.

Completed: 2026-03-12

2026-03-09 16:00 ET — Study Group landing page — card titles for “Who Tends to Find This Useful”

Section titled “2026-03-09 16:00 ET — Study Group landing page — card titles for “Who Tends to Find This Useful””

Added by Claude Code on Patrick’s Mac

Document: spring-2026-landing-page-copy

What was added: Uppercase card titles for the “Who Tends to Find This Useful” section, matching the visual pattern of the “Is This Right For You?” cards (PERSONAL CONDITION / TIME COMMITMENT / PROGRAM OBJECTIVES).

CardTitle
Card 1 (active people, gap between capability and understanding)PHYSICAL BACKGROUND
Card 2 (corrections that didn’t stick, curiosity about mechanics)COMMON EXPERIENCE
Card 3 (educators and professionals)MOVEMENT PROFESSIONALS
Card 4 (willingness to study)WHAT MATTERS

Titles are marked in red in the live copy section of the document. They apply regardless of which approach (A, B, or C) is chosen for the card body text.

Requested action: Apply these titles in Elementor on the live page, using the same styling as the “Is This Right For You?” card titles.

Completed: 2026-03-09

2026-03-09 15:00 ET — Terms of Service updated — Event Validity Periods section + Quebec e-commerce compliance research

Section titled “2026-03-09 15:00 ET — Terms of Service updated — Event Validity Periods section + Quebec e-commerce compliance research”

Added by Claude Code on Patrick’s Mac

What was done (3 items):

1. New “Event Validity Periods and Pass Expiry” section added to Terms of Service (EN + FR)

A general clause added under “Event Participation Terms” covering any event where passes, credits, or tickets are sold for a defined period. Key provisions:

  • Passes/credits valid only within the stated Event Period
  • Automatic expiry at end of period, no refund
  • Cancellation/rebooking subject to each event’s own cancellation window
  • No cross-period transfers under any circumstances
  • Capacity not guaranteed by purchase

Status:

  • French TOS (live): Updated on WordPress (ID 48334), timestamp → 9 mars 2026.
  • English TOS (vault only): New section added in red text. NOT pushed to live site — English terms are managed by Termageddon.

Vault files updated:

  • 02-areas/website/legal/current-policies/terms-of-service.md
  • 02-areas/website/legal/french-policies/terms-of-service-fr.md
  • 02-areas/website/legal/policy-changelog.md

2. Quebec e-commerce compliance research added to compliance roadmap

Key finding: French policies/terms alone are likely not sufficient for full Bill 96 compliance. Baseworks has strong mitigating factors (adult education exemption, good-faith compliance effort, small team). Recommendation: prioritize Phase 2 Quebec lawyer review.

3. Practice Sessions T&Cs addendum context

The new general “Event Validity Periods” clause provides the umbrella language. Decision still needed on credit expiry mechanism (Option 1/2/3) before the PS addendum can be finalized.

Outstanding item: English TOS website update (adding Event Validity section to the live Termageddon-managed page).

Completed: 2026-03-09

2026-03-08 23:30 ET — Study Group landing page — Round 2 voice review + voice guide updates

Section titled “2026-03-08 23:30 ET — Study Group landing page — Round 2 voice review + voice guide updates”

Added by Claude Code on Patrick’s Mac

Document: spring-2026-landing-page-copy

What was done: Patrick reviewed the Spring 2026 Study Group landing page copy against the voice guides. 12 proposed edits added (in a “Proposed Edits — Round 2” section).

Voice guide updates (VOICE-GUIDE-UNIFIED.md):

  • v1.2 draft: “Allowed Metaphors” section added. First entry: “reverse engineered.”
  • v1.3 draft: “Replacing ‘fitness’ in copy” section added — full replacement table. Asia’s addition of “athletic level” incorporated.

Asia’s comment: all study group changes are implemented. voice guide updates reviewed.

Completed: 2026-03-09

2026-03-08 21:45 ET — Backup scripts were broken — now fixed; Fathom Analytics removed; Google Fonts fixed

Section titled “2026-03-08 21:45 ET — Backup scripts were broken — now fixed; Fathom Analytics removed; Google Fonts fixed”

Added by Claude Code on Patrick’s Mac

Important: All 6 automated backup scripts (cron jobs) had never worked. Fixed 2026-03-08. Two changes per script: full path to b2 CLI; environment variable auth instead of b2 account authorize. All tested successfully.

Also done:

  • Fathom Analytics removed from baseworks.com (4 orphaned DB options + 4 draft Fluent Snippet files deleted)
  • Google Fonts self-hosted on baseworks.com (Roboto, Roboto Slab)
  • Google Fonts self-hosted on practice.baseworks.com (Poppins via BuddyBoss/Redux)

Completed: 2026-03-09

2026-03-08 04:00 ET — French Privacy Policy regenerated with Termageddon updates

Section titled “2026-03-08 04:00 ET — French Privacy Policy regenerated with Termageddon updates”

Added by Claude Code on VPS

What happened: The French Privacy Policy page on baseworks.com was regenerated from the updated vault source (privacy-policy-fr.md).

Changes: Added Publicité et remarketing + Sécurité et prévention du spam sections; expanded Programmes d’analytique; all bare URLs now clickable hyperlinks.

Completed: 2026-03-09

2026-03-08 01:45 ET — Quebec compliance Phase 1a complete — all French policy pages live, cross-links added, Phase 1b/1c ready for implementation

Section titled “2026-03-08 01:45 ET — Quebec compliance Phase 1a complete — all French policy pages live, cross-links added, Phase 1b/1c ready for implementation”

Added by Claude Code on VPS

Phase 1a complete: All 5 French policy pages live on baseworks.com; all 5 English policy pages have “Version française disponible” cross-links.

5 French policy pages live:

French pageURLWP ID
Politique de confidentialité/politique-de-confidentialite/48335
Politique relative aux cookies/politique-de-cookies/48341
Contrat de licence (CLUF)/contrat-de-licence/48342
Avis de non-responsabilité/avis-de-non-responsabilite/48343
Conditions d’utilisation/conditions-dutilisation/48334

Phase 1b (checkout checkbox) and Phase 1c (French practice page) were ready for implementation.

Asia’s note: reviewed everything. Also, implemented a bilingual checkbox at checkout. Since implementing a checkbox at the study group page will require using a different page element / custom code, proposed to Patrick that no checkbox is added to the study group “Join the Program” button (moreover, the text “By purchasing any event or online offerings, you agree to our participation terms.” does appear below the button). Potentially, add a checkbox such as “I understand the program will be in English and that by joining the program, I agree to the terms” to the Intake form where it redirects to purchase the Intro session.

Decision forwarded to Patrick’s inbox — 2026-03-09.

Completed: 2026-03-09

2026-03-07 22:00 ET — Privacy Policy updated — analytics, advertising, and security sections added

Section titled “2026-03-07 22:00 ET — Privacy Policy updated — analytics, advertising, and security sections added”

Added by Claude Code on Patrick’s Mac

Context: Termageddon regenerated the Privacy Policy with several new/expanded sections. The vault copies (English + French) have been updated to match.

Changes made:

  • Analytics Programs — expanded: now includes Google Analytics (with opt-out link) + Fathom Analytics + Meta Analytics (previously only Meta Analytics)
  • Advertising and Remarketing — new section: Meta Pixel disclosure
  • Security and Spam Prevention — new section: reCAPTCHA + Cloudflare Turnstile disclosures
  • Date updated to Mar 07, 2026

Files updated:

FileChange
current-policies/privacy-policy.md3 section updates, date
french-policies/privacy-policy-fr.mdMatching French translations, date
policy-changelog.mdNew changelog entry

Action: FYI only — no action needed. The French privacy policy WordPress page (WP ID 48335) will need its content updated to match the new French vault copy when convenient.

Completed: 2026-03-08

2026-03-07 19:15 ET — KB deploy pipeline fixes and frontmatter rules

Section titled “2026-03-07 19:15 ET — KB deploy pipeline fixes and frontmatter rules”

Added by Claude Code on Patrick’s Mac

What happened: The KB site (kb.baseworks.com) deploy pipeline was failing for ~36 hours (March 5–7). Two issues found and fixed: duplicate vault_path YAML keys; broken frontmatter in quebec-language-compliance.md.

New safeguards added: YAML auto-repair; concurrency control; Slack failure alerts (replacing email).

Frontmatter rules added to vault CLAUDE.md: Always use --- delimiters; no markdown syntax inside frontmatter; no vault_path in source files; no duplicate YAML keys; quote titles with special characters.

Completed: 2026-03-09

2026-03-05 20:30 ET — Practice Sessions page — Round 3 complete, ready for Elementor deployment

Section titled “2026-03-05 20:30 ET — Practice Sessions page — Round 3 complete, ready for Elementor deployment”

Added by Claude Code on Patrick’s Mac

Document: practice-sessions-montreal-2026-page-draft-and-form-discussion

What’s done: Intake form fully redesigned (v2); all step questions revised; all outcome card text revised; new section title “Find the right starting point”; new hero CTA “Find Your Starting Point”; community section finalized; FAQ revisions (4 entries + 2 new); voice guide v1.1 draft.

PHP snippet ready at: baseworks-changelog/sites/baseworks.com/code-snippets/2026-03-05-practice-sessions-quiz-v2.php

Completed: 2026-03-09

2026-03-05 15:30 ET — Complete backup infrastructure deployed across all Baseworks sites + n8n server

Section titled “2026-03-05 15:30 ET — Complete backup infrastructure deployed across all Baseworks sites + n8n server”

Added by Claude Code on baseworks-agents VPS

What’s running:

Schedule (ET)WhatWhere stored
Daily 3:30amn8n serverB2: baseworks-n8n-server
Sunday 4:00ambaseworks.comB2: cbbaseworksite
Sunday 4:15amstaging.baseworks.comB2: xCloud1
Sunday 4:30ampractice.baseworks.comB2: cbbpracticesite
Sunday 4:45amcrm.baseworks.comB2: cbcrmbaseworks
Sunday 5:00amVPS configsB2: baseworks-n8n-server

All scripts at /home/patrick/scripts/ on the baseworks-agents VPS. 30-day retention on all buckets.

Completed: 2026-03-09

2026-03-05 12:00 ET — v2 plugin changes documented, ready to reapply after restore

Section titled “2026-03-05 12:00 ET — v2 plugin changes documented, ready to reapply after restore”

Added by Claude Code on Asia’s Mac Mini

Context: Session 8 (2026-03-05) implemented v2 plugins and ran plugin updates. Pre-existing performance issue (~32s cold PHP response) discovered. Decision to revert to 5:45am backup.

Preserved:

  • Plugin files (v2): /Users/asiahome/baseworks-plugin-backups/baseworks-practice-sessions.php and baseworks-session-booking.php
  • Full changelog: /Users/asiahome/Documents/baseworks-changelog/changelog-entries/20260305-120000-practice-sessions-v2-and-plugin-updates.md

Completed: 2026-03-09

2026-03-04 23:15 ET — Practice Sessions page — Voice guide alignment round 2 (full page review)

Section titled “2026-03-04 23:15 ET — Practice Sessions page — Voice guide alignment round 2 (full page review)”

Added by Claude Code on Patrick’s Mac

Document: practice-sessions-montreal-2026-page-draft-and-form-discussion

Summary of changes proposed: Hero subhead; Format section (“theory in context”); Track A & B titles/subtitles/paragraphs; “A note on experience” callout; Inside the Session (3 cards); Pricing; Schedule; Intake form; Booking widget; Community section; FAQ (4 entries); Program ecosystem; Closing CTA.

HTML code produced (for Elementor): Standard pricing table, Alumni pricing table, March–June schedule tables, Proto Studio venue section, Community enrollment box.

Completed: 2026-03-09

2026-03-04 15:30 ET — Automation infrastructure notes — first draft ready for review

Section titled “2026-03-04 15:30 ET — Automation infrastructure notes — first draft ready for review”

Added by Claude Code on Asia’s Mac Mini

Documents:

Summary: All three notes drafted from SSH data. Primer deadline recipe (41083) fully documented. Dual-product pattern confirmed. Spring Study Group gap documented.

Open question that was raised: Primer trigger unification (replace 6 individual product triggers with single WCPURCHPRODUCTINTAG trigger on “Primer” WC product tag). Sub-question: Study Labs recipe consolidation.

Asia’s response: the study labs will also follow the three-month recipe so I deactivated the two-month recipe. I will return to this practically when we start planning for any study labs.

Completed: 2026-03-09

2026-03-03 14:30 ET — Set up QMD semantic search on your Mac

Section titled “2026-03-03 14:30 ET — Set up QMD semantic search on your Mac”

Added by Claude Code on one of Patrick’s machines (predates per-machine attribution)

Document: QMD-Semantic-Search-Setup Requested action: Follow the step-by-step instructions below and in the linked doc. Total time: ~15 minutes (mostly waiting for the initial embedding build).

What is QMD? A local semantic search tool that indexes the Obsidian vault into vectors. Instead of keyword matching, it understands meaning — so searching “how do practitioners progress through forms” finds relevant content even if those exact words don’t appear together. It integrates with both Claude Code and Claude Desktop via MCP.

Already done (Patrick’s Mac + VPS): QMD is running on Patrick’s Mac (1,150 files indexed) and the VPS. Nightly reindex is automated on all machines.

Steps for your Mac:

  1. Install QMD:

    Terminal window
    npm install -g @tobilu/qmd
    qmd --version   # should show 1.0.7 or later

  2. Create the collection:

    Terminal window
    qmd add baseworks-kb ~/Obsidian/baseworks-kb-shared-brain/ --pattern "**/*.md"

  3. Add context description:

    Terminal window
    qmd context baseworks-kb / "Baseworks knowledge base: movement methodology, educational programs, practice platform, Forms, Movement Principles, voice guides, agent system documentation, session summaries, and organizational notes for Patrick and Asia."

  4. Build the initial index (downloads ~2 GB of models on first run, then indexes all files — 5–15 min):

    Terminal window
    qmd embed

  5. Test it:

    Terminal window
    qmd search "what are Movement Principles" -c baseworks-kb

  6. Add QMD to Claude Code — edit ~/.claude/settings.json:

    {
      "mcpServers": {
        "qmd": {
          "command": "/opt/homebrew/bin/qmd",
          "args": ["mcp"]
        }
      }
    }

    If the file already has other entries under mcpServers, add qmd alongside them. Note: Check which qmd — if the path is different from /opt/homebrew/bin/qmd, use that path instead.

  7. Add QMD to Claude Desktop — edit ~/Library/Application Support/Claude/claude_desktop_config.json. First create a wrapper script to prevent terminal noise:

    Create ~/.claude/mcp-servers/qmd-mcp.sh:

    #!/bin/bash
    export NO_COLOR=1
    export TERM=dumb
    exec /opt/homebrew/bin/qmd mcp 2>/dev/null

    Then chmod +x ~/.claude/mcp-servers/qmd-mcp.sh and add to the Claude Desktop config:

    "qmd": {
      "command": "/Users/asia/.claude/mcp-servers/qmd-mcp.sh",
      "args": []
    }

    (Adjust the path to match your home directory.)

  8. Set up nightly reindex — create ~/Library/LaunchAgents/com.baseworks.qmd-reindex.plist:

    <?xml version="1.0" encoding="UTF-8"?>
    <!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
    <plist version="1.0">
    <dict>
        <key>Label</key>
        <string>com.baseworks.qmd-reindex</string>
        <key>ProgramArguments</key>
        <array>
            <string>/opt/homebrew/bin/qmd</string>
            <string>embed</string>
        </array>
        <key>StartCalendarInterval</key>
        <dict>
            <key>Hour</key>
            <integer>3</integer>
            <key>Minute</key>
            <integer>0</integer>
        </dict>
        <key>StandardOutPath</key>
        <string>/tmp/qmd-reindex.log</string>
        <key>StandardErrorPath</key>
        <string>/tmp/qmd-reindex-error.log</string>
    </dict>
    </plist>

    Then load it:

    Terminal window
    launchctl load ~/Library/LaunchAgents/com.baseworks.qmd-reindex.plist

    This runs at 3 AM nightly. If your Mac is asleep, it catches up on next wake.

  9. Verify:

    Terminal window
    launchctl list | grep qmd
    qmd status

Full reference doc: QMD-Semantic-Search-Setup

Completed: 2026-03-03

2026-03-03 14:30 ET — Complete Claude Code first-run wizard on VPS

Section titled “2026-03-03 14:30 ET — Complete Claude Code first-run wizard on VPS”

Added by Claude Code on one of Patrick’s machines (predates per-machine attribution)

Document: Claude-Code-VPS-Login Requested action: SSH into the VPS and run through the Claude Code first login. Your account is ready — just needs authentication.

Steps:

  1. SSH in: ssh asia@46.224.129.16
  2. Attach to the tmux session: tmux attach -t kb (or any session)
  3. Run: claude
  4. Follow the first-run wizard — it will ask you to authenticate with your Claude account
  5. Once done, Claude Code is ready to use on the VPS

Completed: 2026-03-04

2026-03-03 16:30 ET — Inbox notification test

Section titled “2026-03-03 16:30 ET — Inbox notification test”

Added by Claude Code on one of Asia’s machines (predates per-machine attribution)

Document: (test) Requested action: Confirm you received a Slack DM for this item Summary: Testing the automated inbox notification workflow. If you see this in Slack, it works.

Completed: 2026-03-03