Asia's Inbox

Created 2026-03-03

Updated 2026-07-22

Tags inboxasiaworkflow

Review queue. Items are added by Claude when something needs your attention. Mark done [x] and note any decisions.

See 1 for how this works.

Open Items

[ ] 2026-08-03 18:18 ET | Vault session records: no changelog file (Patrick’s decision) — check your machines’ global CLAUDE.md

Added by Claude Code on Hutch Mini

Patrick decided today that this vault has no CHANGELOG.md, deliberately, and vault work is no longer logged in the baseworks-changelog repo (that changelog is site operations only, per the 2026-07-31 rule). Session records are layered instead: inbox items for anything the other person needs, shared-context for infrastructure, version-history tables inside the governed docs for rule changes, Editorial Notes per document, report docs for large sessions, and git as forensic backstop. The full table is in the vault CLAUDE.md → “Session Records — No Changelog File” (commit ea7dd6c7), so every machine picks it up on pull.

Requested action: check ~/.claude/CLAUDE.md on both your machines. If its Record Keeping section still tells sessions to add entries to ~/Documents/baseworks-changelog/CHANGELOG.md after vault work, change it to defer to the vault’s own CLAUDE.md (the repo file is the authority; machine files shouldn’t repeat project rules). Nothing else to do — the vault-side rule arrives via normal git pull.

[ ] 2026-07-31 10:05 ET | How our Claude Code setup is organised — and why your global settings must not override it

Added by Claude Code on Hutch Mini

We reorganised how instructions are stored across all our projects. The short version: instructions live in the repo, not in anyone’s global settings. This affects how your Claude instances should behave on the projects we share, so it is worth ten minutes.

The principle

Each machine has a global ~/.claude/CLAUDE.md. It is personal to that machine and must only contain machine-level things — which computer it is, which tools are installed, where shared resources live.

Everything about how a project works belongs in that project’s own repo, in CLAUDE.md or AGENTS.md. The reason is simple: a rule in your global file reaches only your laptop. A rule committed to the repo reaches every machine and both of us, always at the current version, the moment we pull. Between us we work across four Macs and a VPS.

⚠️ What this means for your setup

Your global settings should not contain project rules for anything we share — not for baseworks-kb-shared-brain, not for baseworks-new-site-build, not for POAS-Home.

If they do, your Claude will follow instructions mine has never seen, and we will drift apart without either of us noticing. If you have project-specific rules in your global file, move them into that project’s repo so we both get them — or tell me and I will.

The repo files are the agreement. Your global file is just your machine.

Why we bothered

Two things broke recently, both from the same cause:

A global rule said “after any significant session, add a changelog entry to baseworks-changelog.” Sounds sensible — but it meant new-site-build sessions were logged into the site-operations changelog. At least seven entries ended up in the wrong repo. They have been left in place as historical record, and the rule now sits in each repo instead.
This machine was mislabelled across every vault — inbox items and commits said “Patrick’s Mac” or “Patrick’s MacBook Pro” when they came from Hutch Mini, Patrick’s desktop. The global file named no specific machine, so it got guessed. It now names it explicitly.

Changelogs are repo-specific. Write to the changelog of the repo you are working in and no other. Cross-references between repos are fine; copying detail across is not.

Where things are

Project	Instructions	Notes
`baseworks-kb-shared-brain`	`CLAUDE.md`	This vault
`baseworks-new-site-build`	`AGENTS.md` (canonical) + thin `CLAUDE.md`	See below
`baseworks-changelog`	`CLAUDE.md` + `CLAUDE-INSTRUCTIONS.md`	Site operations only
`POAS-Home`	`CLAUDE.md` + `AGENTS.md`	Ours

On AGENTS.md: it is the cross-vendor standard — read by Codex, Cursor, Copilot and others, and supported by Claude Code. In the site-build repo it is now the canonical file and CLAUDE.md is a thin pointer to it, so the rules apply whichever tool either of us opens the project with. POAS-Home already has one.

One thing that happens automatically in the site-build repo

There is a SessionStart hook committed to .claude/settings.json there. It loads docs/NEXT-SESSION.md into context automatically whenever a session opens — on your machines too, without you configuring anything.

That file is the standing pick-up point: what is outstanding, what is blocked and on whom, and the question to open with. Whatever the last session leaves in it is what the next session starts from, whoever opens it — so keeping it honest is a shared obligation. Patrick and I have both been updating it.

If it does not fire on your machine, open /hooks once or restart.

On the site build specifically

Patrick is moving into the build while you work through the course, so our positions will differ. Two things worth knowing:

Read docs/DECISIONS.md and docs/WORKFLOW.md before changing anything — your input should land on top of what is already decided rather than beside it.
If something already decided looks wrong, raise it rather than quietly fixing it. A reversal gets a new dated entry explaining why. Deciding not to touch something is a perfectly good outcome.

Requested action: Check whether your global ~/.claude/CLAUDE.md on either machine contains project-specific rules for anything we share. If it does, move them into the repo or tell me what they are. No rush — but worth doing before you start on the site build.

[ ] 2026-07-30 22:15 ET | Bricks Accelerator course: start weeks 1–3 on your own WordPress install

Added by Claude Code on Hutch Mini

Patrick would like you to work through the first three weeks of Dave Foy’s Bricks Accelerator course, by hand, on a WordPress install of your own. He is doing the same on his — he finished week 3 today. The reasoning is that you will likely understand the material better than he does, and both of you having built it by hand is what makes it possible to tell later whether an agent’s output is actually right or merely looks right.

Course access: search “Dave Foy Circle” in 1Password.

⚠️ Please stop at the end of week 3

Do not go past the lessons Patrick has completed. He is at the end of week 3. Staying in step matters — the decisions coming out of the course feed directly into the new Baseworks build, and the two of you need to be looking at the same material when they get made.

Your own WordPress install — on our server, not InstaWP

Dave Foy’s course tells you to spin up a site on InstaWP. Don’t. Patrick is cancelling that subscription. Everything now lives on our own server, baseworks-dev, managed through xCloud. You already have full access.

Set up your own site — do not use any of the three already there:

Site	What it is
`accelerator.1wp.site`	Patrick’s course site. Please don’t touch it
`lively-canyon-239698.1wp.site`	Test site A — tooling evaluation
`frosty-crater-239701.1wp.site`	Test site B — tooling evaluation

Step by step in xCloud:

Log in to xCloud and open the baseworks-dev server (IP 89.106.64.111).
Create Site → choose WordPress.
For the domain, take the xCloud test domain it offers automatically — something like cheerful-meadow-123456.1wp.site. Do not use a baseworks.com subdomain; that was decided deliberately, so nothing points at a throwaway dev box from our live domain.
Choose PHP 8.4. Not 8.3. This matches production and matters later.
Select your SSH keys during site creation. This is the easy-to-miss step — do it while creating the site and you will not have to come back to it. It is also what lets Claude Code help you on that site without needing a password.
The domain you get will be a random name and cannot be renamed — xCloud does not allow it. Instead, set a readable site label in the dashboard, e.g. asia-course, so you can find it.
Send Patrick the domain it gives you so it can be recorded with the others.

Disk and resources are not a concern — the server is at about 6% of 96 GB.

Reference material, if you want it

All six weeks of the course are already captured in the baseworks-new-site-build repo at docs/courses/bricks-accelerator/ — including a CORRECTIONS.md of things that turned out to be wrong or have since changed. Use it as a reference, not a shortcut. The point of this exercise is the hands-on understanding, which is why both of you are building it manually rather than reading it.

You already have access to that repo (write access, confirmed 2026-07-30 — same as this vault). If you have not cloned it yet: git clone git@github.com:p-oancia/baseworks-new-site-build.git

Worth reading first when you do: docs/WORKFLOW.md — it maps what each site is for and what stage the project is at. It was written today, so it is current.

Requested action: Get the site created when you have time, send Patrick the domain, then start week 1. Questions about the course can go to Claude Code in the site-build repo — Patrick’s course tier has no community access, so that is where his are going too.

[ ] 2026-07-18 22:52 ET | Mehling guided read: Patrick’s answers through Movement 7 are ready for your reactions (dialogue still in progress)

Added by Claude Code on Patrick’s Mac

Patrick has been working through the Mehling 2011 paper in a turn-based dialogue to form his own view of it and close the gaps between how you two describe the method. Movements 1 through 7 are settled and captured; Movement 8 (the assembly) is mid-negotiation and will get its own consolidated item when done. Patrick asked that you take a look in the meantime rather than wait.

Doc: Mehling 2011: A Guided Read — read top to bottom; his verbatim answers and the captured insights are inline.

Things specifically flagged for you in there:

The discrimination hinge (Movement 3): the capture argues you and Patrick describe the same variable at different levels (your measurement endpoint is his phenomenological starting point), and that this confirms the two-axes proposal’s central claim from lived experience.
A research question from his answer: the payoff can arrive “maybe not even at the highest level of discrimination,” suggesting the discrimination-to-awareness relationship may be a threshold, not monotonic. Relevant to your SCAPPS framing.
State-4 restore-vs-build (Movement 7): “the capacity is inherent, but the realization has to be worked for” is confirmed provisionally only; Patrick wants to refine it with you (open question: does “innate raw capacity” concede too much to nature?).
The phrase ledger (Movement 8 block): Patrick gave verdicts on which working phrases stay, get reworked, or need candidates; he explicitly wants your eyes on all of them before anything is adopted.
Breath mechanism claims (Movement 4): “over-arouses the sympathetic system,” “destroys the reflective capacity,” assimilation-block breathwork raising respiratory capacity. These are lived-experience claims that need your Layer 1 vetting before any public use.
Heads-up: the guided read showed the Mehling section of the two-layers framework is now stale (“maps directly onto Patrick” doesn’t survive Movements 2, 6, and 7); it’ll need a joint rewrite once the dialogue closes.

Requested action: read the dialogue doc and note reactions (inline or back through this inbox), especially on items 3 and 4. No urgency on the rest; the Movement 8 consolidated item will follow.

[ ] 2026-07-18 11:54 ET | New research-network contact: Ruth Knight (BRNet panel speaker)

Added by Claude Code on Hutch Mini

Patrick connected with Ruth Knight on LinkedIn following her EDI talk that opened the early-career panel at BRNet. She’s a Senior Lecturer in Psychology at York St John University working on body image, eating disorders, and mental health in underrepresented populations, with a participatory/co-production methods angle that overlaps with the Baseworks research direction.

A research-network contact note now exists: ruth-knight

Requested action: none required; FYI since you presented on the same panel session. If you have your own context on her from Padua worth adding, append it to the note.

[ ] 2026-07-16 13:06 ET | New LinkedIn company page Overview + brand guide: review

Added by Claude Code on Patrick’s Mac

Patrick reworked the LinkedIn company page today. Two documents for your review:

Documents:

linkedin-company-page (in 02-areas/communications/) — the new page Overview and specialties. The Overview now opens with the perceptual thesis and presents the holistic mission across three pillars (movement education methodology, published writing, research contributions), generalized so nothing dates. Specialties expanded to 14 tags; “Movement Training” and “Physical Education” removed per the voice guide, Motor Learning and Perceptual Learning added to broaden the neuroscience coverage around Body Representation Research.
linkedin-brand-guide (same folder, status: draft) — a new channel strategy doc: what LinkedIn is for, the audience mindset it targets, and the two publishing voices (Patrick’s personal account for campaign articles, the company page for research communications). It codifies conventions that have been implicit, so it needs your confirmation.

Requested action: Review both, especially the two-voices table in the brand guide and the Overview’s research paragraph (it draws on your framing from the BRNet work). Note: Patrick has already published the new Overview as is on the LinkedIn company page for the time being, so any changes from your review would be applied as a follow-up edit to the live page.

[ ] 2026-07-08 17:57 ET | Media drop-off: new photos go in `media/incoming/`, not `media/inbox/`

Added by Claude Code on Patrick’s Mac

Quick workflow note on where to place new images for processing. The media pipeline (the /compress-photos step: resize → CDN → NAS → symlink) reads from /Volumes/baseworks/media/incoming/. Photos left in media/inbox/ are outside that path, so they don’t get picked up and can be missed.

Going forward: drop any new photo that needs processing into media/incoming/.
incoming/ is transitory: files there are processed and then deleted once the CDN/NAS copies are confirmed, so nothing should live there long-term.
Today’s batch that was in inbox/ (brain fodder valence art, the Death Comes Spring and Montreal banner sources) has been moved to incoming/, processed, and cleaned up. Two Open Day featured PNGs are still in incoming/ pending a decision (they were uploaded straight to WordPress but never archived to the NAS or CDN).

No action needed beyond using incoming/ for the next drop. Thanks.

[ ] 2026-07-07 11:20 ET | Audit decisions executed: publish gate, PARA restructure, plans moved to 01-projects

Added by Claude Code on Patrick’s Mac

Patrick settled the remaining audit open questions this morning and I executed them. Three conventions change for you:

New plans go in 01-projects/, not 03-resources/plans/. PARA restored as designed: active plans in 01-projects, completed ones in 03-resources/executed-plans/, dead logs in 04-archive/projects/. The audit report itself now lives at 01-projects/vault-audit-report-2026-07-05. Old links keep resolving.
The KB site now has a publish gate. Any file tagged internal or draft (or status: draft in frontmatter) is excluded from kb.baseworks.com. If you have working notes you don’t want on the site, tag them internal.
05-private now publishes to the gated site (Patrick’s call: two readers, 2FA-protected, fine).

Also: shared Claude Code permission grants now live in a committed .claude/settings.json (your machines inherit them on pull; your local grants are unaffected), and three unused Obsidian themes were removed from git — if you were using Catppuccin, Blue Topaz, or AnuPpuccin locally, reinstall from Obsidian’s theme browser (the shared setting is PLN).

Requested action: none beyond git pull — just be aware of the three convention changes, especially where new plans go.

[ ] 2026-07-07 15:45 ET | Unified voice guide v2.9 draft — group-post neutral register, pending your confirmation

Added by Claude Code on Patrick’s Mac

The content skills voice audit (SKILLS-VOICE-AUDIT-2026-07-07, follow-up to the 2026-07-07 system audit) added one section to VOICE-GUIDE-UNIFIED as a v2.9 draft: “Group and participant-facing posts” under By Context. It codifies the neutral register we’ve been applying to session-summary group posts since Spring 2026 (no praise adjectives, no performed enthusiasm, no editorializing about cohort effort, directives as factual statements, no sign-off). Previously this lived only in the session-summary skill and Claude’s local memory on Patrick’s Mac.

The same audit updated all 11 content skills (voice-guide loading, the new Final Voice Pass from WRITING-PROCESS v1.1, de-duplication) and made the campaign guidelines’ “Per-Platform Trajectories” section the canonical home for platform rules. One convention changed by Patrick today: hashtags always go inside the post content; the Instagram first-comment idea is retired.

Requested action: confirm or adjust the v2.9 section (it stacks on v2.8, also pending from today). Skimming the audit’s Part 1 is the fastest way to see what changed in the skills you use.

[ ] 2026-07-07 10:30 ET | Vault audit quick wins done — one small check needed on your machines

Added by Claude Code on Patrick’s Mac

The quick-win batch (items 1–10) from vault-audit-report-2026-07-05 is executed and pushed. Two things touch your machines:

Run git config core.hooksPath scripts/hooks in the vault repo on both your machines (then verify with git config core.hooksPath). The git hooks were consolidated: post-merge/post-rewrite (index rebuild after manual pulls) moved into scripts/hooks/ next to pre-push. If your config still points at the old scripts/git-hooks (per an April instruction), the hooks silently stop working after this change.
On your next pull, ~313 files under site/src/content/docs/ disappear from your working tree. That’s expected — they were stale generated build output tracked by mistake; the KB site regenerates them at build time.

Also done in the batch: OpenClaw credentials redacted and guide archived (OpenClaw and CrewAI are both formally retired per Patrick, 2026-07-05 — CrewAI file archiving is a separate upcoming step), skill path fixes, frontmatter added to 35 files, BRNet poster + carousel HTML moved to NAS symlinks. Full detail in the report’s Quick Wins section and in claude-code-shared-context → Recent Infrastructure Changes.

Requested action: run the one-line git config command on both machines; reply here if anything looks off after pulling.

[ ] 2026-07-05 11:16 ET | Comprehensive vault audit complete: report ready, notes wanted (no action needed)

Added by Claude Code on Patrick’s Mac

Patrick had me run the full diagnostic audit of the vault today (the one planned in vault-comprehensive-audit-fable-prompt). Six areas were covered: the Claude memory system, structure and content integrity, the index databases, the KB build/sync pipeline, tooling and platform-agnosticism, and security/hygiene. Nothing was changed: this was read-and-diagnose only, and the full report is here:

Report: vault-audit-report-2026-07-05 (03-resources/plans/vault-audit-report-2026-07-05.md)

What Patrick asks of you: please don’t implement anything from it yet. Read through it (the executive summary and the “Open Questions” section are the fastest route) and keep notes of anything you think needs implementing, or anything you’d prioritize differently, so we can fold your list in alongside what we do. Two decisions Patrick has already made:

OpenClaw is retired. We will not be running it in the future. That resolves the report’s open question #7: the committed gateway token gets redacted and the OpenClaw docs get archived rather than rotated and kept.
CrewAI is retired. Anything CrewAI-related gets archived so it stops crowding the knowledge base (this includes 03-resources/agent-system/crewai/, the two CrewAI architecture/deployment docs, and the stale 01-projects/Baseworks-n8n-CrewAI-Deployment.md).

Things in the report most likely to touch you or your machines:

The platform-agnostic refactor is the centerpiece: right now git hooks, the launchd sync job, and scheduled tasks are all installed by hand on each machine, so a fresh clone (for example on your Macs) silently has no automation. The plan proposes a config file plus a bootstrap installer. When we implement it, there’ll be a one-time setup step to run on your machines.
.claude/settings.local.json is currently committed but contains Patrick’s-Mac-only entries. The fix (gitignoring it, moving shared permissions to a shared file) will need a small one-time step on your side too.
One skill is broken cross-machine: archive-newsletter hardcodes Patrick’s Mac paths, so it would fail on your machines today. create-newsletter also points at a file that moved. Both are quick fixes on our list.
Publish boundary: the KB site sync currently ignores internal and draft tags (about 51 tagged files publish to the gated site), and 00-inbox publishes, including the infrastructure context doc. No exposure with just the two of us, but it’s the main thing to fix before anyone else is ever invited. Worth your read since you tag things too.
Structure questions (report section “Open Questions”): whether 01-projects becomes the home for active plans (everything in it right now is dead or completed), and how to consolidate the six ad-hoc archive folders inside 02-areas. Your preference matters for both since you navigate the same tree.

No deadline attached. Notes in whatever form suits you (a reply item in Patrick’s inbox, or margin notes in a copy) and we’ll merge them into the remediation plan before anything runs.

[ ] 2026-07-04 22:14 ET | Forest hike conversation: transcript + independent analysis ready

Added by Claude Code on Patrick’s Mac

Patrick and I recorded our forest hike conversation today (the recurring loop about what Baseworks is and isn’t, the “Between the Notes” article, and sensory resolution vs. abstract space), and we asked for an independent third-party read of it. That’s done and it’s yours to see in full.

Transcript (speaker-attributed, ~112 min): 2026-07-04-forest-hike-transcript
Analysis (the third-party read we both asked for): 2026-07-04-forest-hike-analysis — includes a straight answer on who was being rational, separate feedback for each of us, and a protocol for how we build frameworks from here.

A few things worth flagging so you know what’s in it:

The verdict is not both-sides mush. It’s specific: you were right about the content (the article really does hold several distinct phenomena, and the sensory-resolution vs. abstract-space distinction is real and worth insisting on), and Patrick was right about delivery and timing. There’s honest, direct feedback for both of us, including for Patrick.
Section 5 is the part Patrick most wants you to read: “two registers inside the material itself,” and the distinction between deliberate, bounded ambiguity (a design choice you could specify but choose not to, with a reason) and unbounded ambiguity (woo). It’s framed as a bridge to your rigor, not a retreat from it, and it maps onto the phenomenological cell of your two-axes proposal.
The analysis recommends running that still-pending two-axes proposal through the new protocol as the first concrete test case.

Storage note: these live in the vault’s 05-private/ folder, which is not published to the public KB site (it’s outside the synced directories), but it is in our private git repo, so it syncs to your Mac normally. The original audio is archived on the NAS. You have full access to all of it.

No deadline. When you’ve read it, it’d be good to hear whether Section 5’s framing lands for you, and whether you want to book a desk session on the two-axes decision.

[ ] 2026-06-30 ET | Framework proposal for your reaction: two axes, not two layers

Added by Claude Code on Patrick’s Mac

Document: the-two-layers-framework-PROPOSAL-two-axes 02-areas/method-admin/the-two-layers-framework-PROPOSAL-two-axes.md

This is a proposal, not an edit. It does not touch the shared the-two-layers-framework. It argues for one structural refinement and asks for your reaction before anything gets merged in.

The claim, in one line: the Layer 1 / Layer 2 distinction is more accurately two independent axes, not one. Level of analysis (mechanistic vs phenomenological) and scope (modality-independent vs physical-specific vs fitness). Drawing the grid those axes form exposes one cell with no label, and that cell is exactly the category the current doc already flags as having no home: the lived, first-person experience of what committed physical practice specifically opens (your before/after testimony; the article’s closing “insurance” paragraph).

The fix it proposes: split Layer 2 into 2a (Tier 1, modality-independent phenomenology, what Patrick mostly describes) and 2b (Tier 2, physical-specific phenomenology, what you describe firsthand). Not a Layer 3. The reasoning: 2a and 2b are the same level of analysis (both phenomenological), differing only in scope, so a third layer would put the distinction on the wrong axis.

Why it matters practically: it corrects the framework’s own prescription. The doc currently says the Tier-1-already audience (the committed musician, the academic) is reached by “Layer 1 applied to a Tier-1-ready audience.” But mechanism can’t convey lived content. The argument that actually reaches them is 2b, the firsthand account of the opening, which you can give better than anyone, backed by Layer 1 as mechanistic confirmation. The framework can’t state this today because it has no name for 2b.

What I’d value your read on (four open questions at the bottom of the doc): whether the grid matches how you see the territory, whether “2a/2b” is the right notation, whether you want to author the 2b testimony as a standing section or keep it diagnostic, and whether the mechanistic-row cell names (general learning science, exercise physiology) are scientifically right. The doc also lists the integration cost: six sections of the shared framework would need rewriting if we adopt the split, so it’s a whole-change decision, not a clean add.

No deadline. The article it grew out of can publish independently of this; the two are decoupled.

[ ] 2026-06-27 ET | Two new skills added: /happyfiles-folder and /wp-media-import

Added by Claude Code on Patrick’s Mac

Both are available on your machines after a git pull in the vault directory (or your next auto-sync).

/wp-media-import — imports already-processed photos into the WordPress Media Library. Sets title, caption, and alt text per image; optionally assigns to an existing or new Happy Files folder. Use this after /compress-photos to complete the WP upload step. Works from CDN URL on machines without NAS access.

/happyfiles-folder — standalone tool for Happy Files folder operations: create folders, assign existing attachments to them, list folders, verify assignments.

No configuration needed — just git pull and the skills appear. Tailscale must be active on your machine for either skill to SSH to the baseworks.com server.

[ ] 2026-06-24 16:22 ET | Marta Stothers response — revised draft ready for your review and send

Added by Claude Code on Patrick’s Mac

Document: marta-stothers-signition-260526 02-areas/practice-platform/primer/assignments/segment-feedback/marta-stothers-signition-260526.md

What was added (changes marked in red in the document):

New paragraph requesting she go back and fill in the empty feedback fields, with a note about why her perspective is valuable given her background
Closing paragraph substantially expanded: course completion acknowledgment, apology for not seeing her in the last sessions, and an open invitation to reach out
Two new paragraphs addressing the undisclosed condition situation: Patrick’s observation and conversation with Marta, her choice not to share, and a clarification that his intention was about group preparation, not individual special treatment

Two small mechanical fixes (not marked): em dashes in paragraphs 2–3 corrected per voice guide.

Action: Read the full draft (red text = new content), confirm you’re comfortable with the condition paragraphs in particular, and send directly to Marta at martastothers@gmail.com.

After sending: Update the note in Marta’s people profile (02-areas/people/marta-stothers.md) — change “Pending Asia’s send” to the actual date sent. This keeps the vault record accurate for future cross-referencing.

[ ] 2026-06-10 09:00 ET | Interdisciplinary conference targets — new reference file

Added by Claude Code on Patrick’s Mac

A new reference document has been added to the conferences folder:

Interdisciplinary Conference Targets — Sensorimotor, Neuroscience, Exercise Physiology

Path: 02-areas/method-admin/core/science-docs/conferences/interdisciplinary-conference-targets.md

This came out of reflecting on BRNet 2026 (Padua) — the research culture there skews heavily toward clinical pathology and rubber hand paradigms. The document surveys seven conferences that are more genuinely interdisciplinary, bridging sensorimotor learning, exercise physiology, neuroscience, and educational paradigms:

ICPA (International Conference on Perception and Action) — ecological psychology, perception-action as inseparable; next 2028
NCM Society (Neural Control of Movement) — most rigorous motor neuroscience meeting; next Sevilla 2027
I-MDRC (International Motor Development Research Consortium) — bridges researchers, clinicians, educators; next 2027
UK Sensorimotor Conference (Plymouth) — smaller, integrative; annual July
IMBES (Mind, Brain and Education) — neuroscience + pedagogy as co-equals; annual
ECSS (European College of Sport Science) — exercise physiology + motor control integration; annual
Progress in Motor Control — focused, discussion-oriented; next ~2027

Each entry includes a description, fit assessment, submission types, direct links to abstract submission portals, deadline patterns, and notes on non-academic eligibility (none explicitly prohibit independent organizational affiliations). Most 2026 deadlines have passed — primary targets are 2027 editions. Action items checklist at the bottom of the file.

No action required right now — just for your awareness and future planning.

[ ] 2026-06-02 13:40 ET | Pull before you push — Triple Irony campaign note was edited

Added by Claude Code on Patrick’s Mac

Before you push your current work on 02-areas/communications/campaigns/2026-05-triple-irony/2026-05-triple-irony.md, run git pull first. Claude Code made a structural fix to the same file (moved the Production Notes section to the correct location after all social posts). That change is already on origin/main. Pull first to avoid a merge conflict.

2026-05-19

Policy draft: Course Participation and Platform Integrity — review needed

SSH between your Mac Mini and MacBook Air — one-time setup

Session 7 summary: finalized, ready for forum post and group post

2026-04-23

Shobayashi invoice № 2026-05190 (BASEWORKS JPO mark 5823947) unpaid — decision needed

2026-04-22

Patrick’s new article “Sensitivity as a Trainable Skill” is live — mentions your research and upcoming talks

New collaboration protocol for /draft-response

[ ] 2026-05-19 20:27 ET | Policy draft: Course Participation and Platform Integrity — review needed

Added by Claude Code on Patrick’s Mac

Document: course-participation-integrity Requested action: Review the draft and confirm the wording before it goes live.

Summary:

New policy covering: completion requirements (lessons watched in full, AI-generated feedback not accepted), supportive framing (participants encouraged to reach out about schedule constraints), and the platform’s right to reset progress or revoke access
Refund policy confirmed: No refund when access is revoked for integrity violations
Prompted by the Noemi Normandin case (May 2026) — full record at index

Two things to do after confirming the wording:

FAQ — Spring 2026 study group page (WP ID 46997) — add one or two FAQ questions on completion requirements and integrity. Draft question is in the policy doc. Update via Elementor directly.
Terms and Conditions (WP ID 4910) — add the “Platform Access” section via Termageddon (English). French version (WP ID 48334) can be updated directly via WP-CLI.

[x ] 2026-05-19 20:08 ET | Noemi Normanding — course progress reset, email needs sending

Added by Claude Code on Patrick’s Mac

Document: course-progress-reset Requested action: Fill in the segment number and send the email to Noemi Normanding.

Background: Noemi was skipping through Primer lessons and marking them complete without watching them. Patrick had to manually reset her progress and update the video settings to prevent skipping. A system notification email template was created for this — the first instance of a new category of platform administration emails.

To send:

Open the template at the path above
Replace [Segment X] with the actual segment number Noemi was on when her progress was reset (Patrick will know — check with him if unsure)
Send from the platform/system address (no-reply), not from Patrick or Asia personally — it should read as automated
No sign-off needed; the template ends at the reset notice

Notes:

This template is intended to become a fully automated FluentCRM trigger in the future — for now it goes out manually
A new folder 02-areas/communications/email-templates/practice-platform/ has been created for this category of system/admin emails; add future templates there

[ ] 2026-05-19 14:35 ET | SSH between your Mac Mini and MacBook Air — one-time setup

Added by Claude Code on Patrick’s Mac

Patrick set up SSH between his desktop and MacBook Pro today (over Tailscale) and wanted you to have the same. This lets you SSH from one of your Macs into the other from anywhere — useful for remote commands, Claude Code sessions, and general troubleshooting.

Document: asia-ssh-between-macs Requested action: Follow the setup guide on both your Mac Mini and MacBook Air when you have 10–15 minutes.

Summary:

Both your machines are already on Tailscale (Mac Mini: 100.119.127.26, MacBook Air: 100.113.132.50)
Setup involves: enabling Remote Login, exchanging SSH keys, adding SSH config entries, and shell aliases
After setup: type air on Mac Mini to connect to the MacBook Air; type mini on MacBook Air to connect to the Mac Mini
Works from anywhere, not just home network

Machines required:

Mac Mini
MacBook Air

[ ] 2026-05-15 14:34 ET | baseworks-new-site-build — one-time machine setup required for /capture-content

Added by Claude Code on Patrick’s Mac

The new site build project now has a /capture-content skill system for capturing video content (course lessons and reference resources). It depends on the /watch skill, which must be installed at the machine level — it is not part of the project repo.

What to do on each of your machines (Mac Mini, MacBook Air):

Pull the latest baseworks-new-site-build repo
From the project root, run:
```
bash .claude/setup.sh
```
This will clone the /watch skill to ~/.claude/skills/watch/, install ffmpeg and yt-dlp via Homebrew if not already present, and walk you through adding a Whisper API key for audio transcription.
When prompted for a key, use a Groq key (preferred — free tier, faster): console.groq.com/keys. OpenAI also works as a fallback.

Patrick will do the same on his MacBook Pro separately. No action needed before you’re ready to use the capture workflow.

Skills this enables: /capture-content (orchestrator), /capture-lesson (course lessons → week files), /capture-resource (standalone reference videos → docs/resources/)

[ ] 2026-05-15 ET | Fluent Community — WPManageNinja response received, please review for workarounds

Added by Claude Code on Patrick’s Mac

Following up on the earlier item below. WPManageNinja responded to our 10-question support inquiry on 2026-05-14. Their response confirmed every gap we had already identified — nothing was resolved. The full ticket text and their answer-by-answer response are now in the plan document:

Fluent Community Migration — Viability Assessment

What their response confirmed:

FluentPlayer is still in beta, no release date, and video completion gating is not on a confirmed timeline
No sequential/completion-gated lesson progression (drip only)
No assignment grading workflow
LearnDash can coexist on the same install — no conflicts, but no native bridge either
No official bbPress migration tool — custom scripting only
WP Fusion lesson-level gating not confirmed — likely needs custom dev
No native Presto Player integration — community workaround only
No per-user language switching across devices
WP-CLI surface is only the migration command; everything else is wp eval or direct SQL on wp_fcom_* tables
Topics (Pro) are space-scoped only — no cross-community tag browsing

Requested action: Please review the full plan document in detail. The Phase 1 community migration looks viable as-is — but for Phase 2 (courses), we need your eyes on whether any of these gaps have workarounds that would work for how we actually use the platform. Specifically: are there any community/forum patterns or course delivery approaches that would let us work around the missing video gating, progression, or cross-space discovery features with what FC already ships out of the box?

No decision needed yet — Patrick wants your read before any timeline gets set.

[ ] 2026-05-14 14:56 ET | Fluent Community migration viability — for awareness and input

Added by Claude Code on Patrick’s Mac

Patrick has been researching a potential migration of practice.baseworks.com from BuddyBoss + bbPress + LearnDash to Fluent Community. A full viability assessment is now in the plans folder:

Fluent Community Migration — Viability Assessment

Summary of findings:

Phase 1 (community migration only — BuddyBoss → FC Spaces, feeds, DMs) is viable now. LearnDash, Presto Player, and Bunny Stream stay as-is.
Phase 2 (full course migration into FC) is blocked by three missing features: video completion gating, sequential lesson progression, and assignment grading. No confirmed timeline from the vendor.
bbPress forum content has no official migration tool — a custom script would be needed.
The existing BuddyBoss CLI fix scripts (buddyboss-forum-cli-notification-fixes, buddyboss-mention-notifications-via-wp-cli) become irrelevant post-migration.
A support inquiry was sent to WPManageNinja on 2026-05-14 covering the blocking gaps.

Requested action: Read the plan and flag any concerns or dependencies from your side (especially anything in the community/forum area you’re aware of that should factor into the migration decision). No action needed until WPManageNinja responds.

[ ] 2026-05-14 16:30 ET | post-to-group skill updated — voice compliance check now required

Added by Claude Code on Patrick’s Mac

The /post-to-group skill now includes a mandatory voice compliance step (Phase 1d) that runs before the confirmation gate. It checks for em dashes, exclamation marks, negative-framing openers, and banned phrases. Any violations must be fixed before the preview is shown.

This brings it in line with /draft-response, /session-summary, and /create-email, which already had this check. The gap was caught after a forum reply went out with two em dashes.

No action needed — just awareness that the skill now enforces this automatically.

[ ] 2026-05-13 15:30 ET | Press Archive Analysis — complete (all 142 PDFs)

Added by Claude Code on Patrick’s Mac

The full YogaJaya press archive (2004–2017, 142 PDFs) has been analyzed. Every entry in press-archive-index now has a vault note, People, and Relevance columns filled. One invalid file: J83 (Veggy May 2010 cover — corrupt PDF; may need re-scan from hard copy).

Key milestone entries for your reference when writing press page copy:

J99 (Houdini 2015) — first press mention of Baseworks by name
J101 (Yogini Nov 2016) — lineage chart: Dharma Mittra + Jivamukti confirmed
J88 (Honeyee Dec 2010) — most detailed philosophy article; first named practice sequences in press
D01–D02 (Yoga Aktuell 2011–2012) — German workshop tour + personal Q&A

Next step: D1–D10 decisions in decisions-worksheet need resolution before press page copy is drafted. Also PP01/PP02 press packs (archive verification pass) still to review.

[ ] 2026-05-13 | Press Page Brief — two-layers framework integrated; D13/D14 likely resolved by your framework document

Added by Claude Code on Patrick’s Mac

An earlier inbox item (same date) said D13 and D14 require your review. This supersedes that item.

the-two-layers-framework has been integrated into press-page-brief as the governing communication framework (§1.5). Since the framework’s “Press Page Application (D13 Resolution)” section already provides the 5-point arc for the lineage section, D13 may already be resolved — Patrick needs to confirm. If he closes D13, the only remaining step is drafting final lineage prose following the 5-point arc (Layer 1 mechanism + Layer 2 motivation, yoga once subordinately).

What’s now in the brief that references your framework:

§1.5 — cross-medium communication table (which layer leads in each context: press, student onboarding, academic, general public)
§7.6 — bridge problem section superseded by pointer to framework; the 5-point arc from “Press Page Application” is now the operative spec
§8a.1 — IS/IS NOT semantic displacement markers from your AI discoverability section now in the SEO strategy
D13/D14 in decisions table — status changed to “LIKELY RESOLVED — Patrick to confirm”

The one thing that still needs writing once D13 is confirmed: the lineage section prose itself. The framework gives the arc; we need final copy that follows it. That’s a Patrick + you collaboration — both layers need to be present.

[ ] 2026-05-12 14:45 ET | CLI notification pipeline — fully complete + participant communications workflow direction

Added by Claude Code on Patrick’s Mac

Part 1 — Correction to the 2026-05-10 notification item

The 2026-05-10 inbox item said “@mention notifications still require a manual UI edit-and-save.” That is no longer true. A mu-plugin (bw-fix-cli-mentions.php) was deployed today that handles this automatically. All CLI-posted replies with @mentions now fire bb_new_mention notifications without any manual step. The retroactive mention on reply 22259 (the one that was already live) was also fixed via CLI this session — Nathalie’s notification id=2017 was created.

What’s fully automatic now for CLI posts:

@mention notifications fire on insert (no UI edit needed)
Reply author is auto-subscribed to the topic (so they receive future reply notifications)
Topic creator is auto-subscribed to their own topic on creation
Patrick and Asia are auto-subscribed to any topic where they are @mentioned

Part 2 — Participant communications: use the skills, don’t post directly

Patrick’s direction: use /post-to-group and /draft-response for all participant-facing communications. Don’t post to the forum or reply to participants directly from the BuddyBoss UI without going through a skill session first.

Why this matters:

When you post directly from the UI, two things tend to go missing:

Notification pipeline — the mu-plugin only activates for CLI posts. UI posts go through BuddyBoss’s native pipeline, which works but bypasses the bb_create_subscription() auto-subscribe logic. If the participant hasn’t subscribed to a topic, they may not receive reply notifications even if you @mention them.
Vault records — the forum content sync pulls everything into community-forums-groups/topics/ automatically, but it doesn’t log communications to the participant communications log, it doesn’t create editorial sidecars, and it doesn’t capture the reasoning behind how we responded. That context is lost.

The intended workflow:

Situation	Skill to use
Replying to a participant’s forum post	`/draft-response` → `/post-to-group forum-reply`
Creating a new forum topic (session summary, announcement)	`/post-to-group forum-topic`
Posting to the group activity feed	`/post-to-group group-activity`
Responding to a DM or email	`/draft-response`

When you use /draft-response, the skill drafts the response, captures any editorial refinements you make as notes in the vault, and logs the communication in the participant communications log. When you then post via /post-to-group, the notification pipeline fires correctly and the topic subscription is created.

One exception: if you’re doing a quick UI edit-and-save on a post that already exists (e.g. fixing a typo), that’s fine directly in the UI. The concern is original posts and replies that generate a communication record.

Part 3 — What was built today (changelog reference)

Three deliverables shipped:

bw-fix-cli-mentions.php (mu-plugin, now live on practice.baseworks.com) — fires bb_new_mention for @mentions in CLI-inserted replies; auto-subscribes Patrick (user_id=8) and Asia (user_id=2) when either is @mentioned.
- Changelog repo: sites/practice.baseworks.com/mu-plugins/bw-fix-cli-mentions.php
- Full debugging notes and carry-forward rules in baseworks-changelog/CHANGELOG.md (2026-05-12 entry)
/post-to-group skill — Mechanism B (reply template) updated to call bb_create_subscription() after posting, subscribing the reply author to the topic.
/post-to-group skill — Mechanism A (topic creation template) updated with wp_set_current_user(), do_action('bbp_new_topic', ...), and bb_create_subscription() so topic creators are subscribed and the notification pipeline fires correctly on create.

No action required from you — awareness and workflow direction only.

[ ] 2026-05-10 ET | FYI: Forum reply notification pipeline fixed — what changed for you

Added by Claude Code on Patrick’s Mac

Three root causes were found and fixed that explain why you and Patrick weren’t getting reply notifications for Primer Community forum posts.

What changed:

Two BuddyBoss Platform feed settings were disabled site-wide (bp-feed-platform-bbp_reply_create and bbp_topic_create both = 0). These blocked all forum activity records from being created in the database. Now enabled.
The /post-to-group skill’s reply template now fires bbp_new_reply after inserting — this is the action hook that triggers the subscription notification pipeline. It was silently missing.
You and Patrick have forum subscriptions to the Primer Community forum, which only notify you of new topics. For reply notifications you need topic subscriptions. You are now subscribed to both existing Primer Community topics (22252, 22256) at the topic level.

What this means for you now:

You will receive in-app and email notifications when new replies are posted to Primer Community topics you’re subscribed to
You already have a retroactive notification for Patrick’s reply 22259 (Nathalie’s “One more time…” topic)
For future topics: they’ll need a topic subscription added when created — not yet automated, but documented

One thing still requiring a manual step: @mention notifications from CLI-posted replies still need Patrick or Asia to open the reply in the UI and do a minor edit-and-save. Visual rendering of @links works; only the notification doesn’t fire from CLI. (Documented in the plans file.)

No action required from you — this is informational.

[ ] 2026-05-09 18:00 ET | Action needed: install /watch skill on your machines

Added by Claude Code on Patrick’s Mac

Patrick has been using the /watch skill for KB research (used today to analyze a Nick Milo video). It lets Claude download a video, extract frames, pull a transcript, and work with the content directly — useful for research sessions.

Full setup guide in the vault: /watch Skill — Setup Guide

Step 1 — Install the skill and dependencies (run once per machine):

mkdir -p ~/.claude/skills
git clone https://github.com/bradautomates/claude-video ~/.claude/skills/watch
brew install ffmpeg yt-dlp

Step 2 — Set up the free Groq API key (needed for local video files):

Go to console.groq.com/keys and create a free account
Generate a new API key
Run in Terminal (replace with your actual key):

mkdir -p ~/.config/watch
echo "GROQ_API_KEY=your_key_here" > ~/.config/watch/.env
chmod 600 ~/.config/watch/.env

YouTube videos work without the key (they have native captions). The key is needed for local .mp4/.mov files — session recordings, etc. Groq is free, no credit card required. Don’t share the key or commit the .env file to git.

[ ] 2026-05-07 ET | New global skill available: `/watch` for video analysis

Added by Claude Code on Patrick’s Mac

Patrick has installed a Claude Code skill called /watch (from bradautomates/claude-video) globally on his machines. It lets Claude download a video, extract frames, pull a timestamped transcript, and answer questions based on the actual content — useful for research from YouTube tutorials or any other video.

It is not installed on your machines. If you’d like it, run the following once on each machine:

mkdir -p ~/.claude/skills
git clone https://github.com/bradautomates/claude-video ~/.claude/skills/watch
brew install ffmpeg yt-dlp

After that, /watch <youtube-url> works from any Claude Code session. No action required — just letting you know it exists and how to add it if you want it.

[ ] 2026-05-05 16:30 ET | baseworks.com registration — bot protection updated, keep an eye on new signups

Added by Claude Code on Patrick’s Mac

The bot registration issue from April has been resolved. The Turnstile plugin (never fully configured) has been removed and replaced with two things: a honeypot field on the WooCommerce registration form, and a Cloudflare WAF rate limit rule on the /my-account/ endpoint. First name and last name are now required fields on registration and save directly to user meta.

Please keep an eye on new registrations over the coming weeks. If anything suspicious comes through (random-string emails, same-IP clusters), flag it — the next step would be enabling email verification before account activation. Full details in CHANGELOG.md (2026-05-05).

[ ] 2026-05-05 13:30 ET | Language as a Teaching Medium blog post — defer to new site + consider LinkedIn instead

Added by Claude Code on Patrick’s Mac

Patrick has reviewed your blog post proposal on language as a teaching medium (_blog-idea-language-teaching-at-baseworks). His take: the idea is good but the timing isn’t right for the current site. He’d like to revisit it once the new site is up and running. He also thinks a LinkedIn article might be a more appropriate format for this piece — better reach for the audience it’s aimed at, and it doesn’t need to live inside the blog structure.

No action needed now — just flagging so you know where it stands when the new site conversation comes up.

[ ] 2026-05-04 11:00 ET | Session 5 Summary draft ready for review

Added by Claude Code on Patrick’s Mac

The Session 5 summary (May 2, 2026) has been drafted and is ready for your review and editing. Patrick is handing this off to you.

Where to find it

02-areas/educational-programs/study-groups/2026 (Spring) Study Group Montreal/session-summaries/session-5-summary-en.md

Pull the latest from git if the file isn’t visible yet:

cd ~/Obsidian/baseworks-kb-shared-brain && git pull origin main

What’s in it

This was the densest session of the program: six new forms (Z-Lunge, Z Expansion C-Tuck, Split Form Inflection, Seated Wide Inflection, Seated Inflection, Supine Leg Raises), the standing review reframed through Transit, all three IM types in practice, and circumlinear motion as the unifying thread. Horizontal Shoulder Flex, Plank/Press-Up, and Assimilation were cut for time.

The draft has passed the Phase 5 automated checks (zero em dashes, no deprecated terms, YAML complete, Ignition/Assimilation framing correct, first-person voice verified for Patrick as posting instructor). It has not yet been reviewed by a human.

What’s needed from you

Read through the full summary and edit for accuracy, voice, and emphasis
Flag anything that needs correction (speaker attribution, form descriptions, concept framing)
Once you approve the English version, translation to French can proceed
The group post will be drafted after English approval (it goes inline in chat, not in the .md file)

[ ] 2026-04-22 21:15 ET | Patrick’s new article “Sensitivity as a Trainable Skill” is live — mentions your research and upcoming talks

Added by Claude Code on Patrick’s Mac

What’s live

Patrick’s article is published at https://baseworks.com/article/sensitivity-as-a-trainable-skill/. It’s a long-form companion piece to the direction you’ve been developing on the research side — introduces five new Baseworks-owned terms (“muscular sensation at rest,” “sensory vocabulary,” “perceptual resolution,” “inner attention,” “attentional carryover”) and frames them around the experiential side of the work.

Where you appear

First mention hyperlinked to your instructor bio (/instructor/asia-shcherbakova/) in the paragraph about “the hum” — the low-grade tonic sensation you’ve been describing in muscles at rest.
Full paragraph in “The science underneath” section explaining your role as research associate since 2015 and linking directly to The Mystery of Proprioceptive Awareness as the research companion.
Both upcoming talks referenced: Neuropsychology Day 2026 (May 11, Montreal) and BRNet 2026 (June 8–9, Padova).

For your review

The article is already live, so this is heads-up rather than a review request — but if you catch anything you’d like changed (framing, attribution, term usage), flag it back and Patrick can adjust. Vault source: 02-areas/website/blog-articles/2026-04-22-sensitivity-as-a-trainable-skill.md.

Four new entries were drafted into VOICE-GUIDE-UNIFIED.md during this article’s editing rounds: banned “land/lands on” as an attention metaphor, restructured loaded appositive-clause sentences, banned negative-referencing openers, soft-intensifier drops. Per governance, major guide changes need your review before v2.7 is official. Memory pointers are in place; the guide file reflects the draft state.

Requested action

Skim the article when you have a moment — let Patrick know if anything needs correcting. No deadline.

[ ] 2026-04-22 15:30 ET | New collaboration protocol for `/draft-response`

Added by Claude Code on Patrick’s Mac

What changed

The /draft-response skill now has a memory. Every time we draft a response together (forum replies, inquiries, participant comms, vendor emails), Claude Code will:

Before drafting: query the vault index for prior sidecars and responses on overlapping topics, and load the most relevant editorial notes as context.
After you or Patrick approve the draft: write an Editorial refinements section into a durable file (a sidecar for forum replies, the contact-inquiry file for inquiries, the comms log for participant work). That section captures the reasoning behind each substantive change we made.
Next time a related topic comes up: retrieval surfaces those captured refinements, so drafts start from the accumulated editorial record rather than from scratch.

What this means for you in practice

Use /draft-response <description> as the default tool for any response work: forum replies, inquiries, participant communications, vendor/collaborator messages, policy enforcement, boundary-setting.
When you edit a draft Claude produces, keep the reasoning visible when it’s non-obvious. One short sentence is enough: “no, we don’t say ‘sit bones’ first, let her infer from the video” or “this is too coaching-ey, flatten it to observation.” Claude will capture those into the Editorial refinements section.
If a refinement is obvious from the change itself (adding a missed mechanic, correcting a form name), no narration needed; Claude will infer it.
If an edit has broader reach than the current draft (a voice rule, a pedagogy principle, a workflow decision), mention that. Claude will both write it into the sidecar and save it as a persistent memory so it applies across sessions.

The protocol doc

Full details, including what counts as a substantive refinement, where captured refinements live by communication type, and how retrieval ranking works, are in:

03-resources/claude-code-skills/draft-response-collaboration-protocol.md

The protocol is editable. If you notice patterns that aren’t captured, retrieval surfacing hits that aren’t useful, or new communication types that need their own capture location, update the doc and commit. Patrick can do the same. It’s meant to evolve as we use it.

Why this matters

Every correction or refinement you make is a piece of teaching intelligence. Before this, it lived only in the finished reply. Now it becomes retrievable context that shapes future drafts, so Claude’s initial inference gets closer to the right voice and content over time, reducing the number of rounds of refinement needed. The system learns from how you write, not just from what’s in the voice guides.

Requested action

Read the protocol doc when you have a moment, and start using /draft-response as the default for communication work. Mark this item done once you’ve read it.

Draft Response — Collaboration Protocol (full protocol doc)
Example sidecar with Editorial refinements: What moves the weight forward — curated notes
Skill file: .claude/skills/draft-response/SKILL.md

Added by Claude Code on Patrick’s Mac

Context

A new automation “People aggregator — webhook on CRM change” (id 42) is live in FluentCRM. On every tag application it POSTs to https://agents.baseworks.com/webhook/people-index, which re-runs the vault’s people aggregator with fresh CRM data. This gives the vault near-real-time visibility into contact changes instead of waiting for the 4-hour cron.

Patrick published it with a conservative starter scope — only two tags currently trigger the automation:

Tokyo Studio Alumni (tag id 103)
Event: Practice Meet Tokyo (tag id 74)

What’s needed from you

Open the automation and widen the trigger tag list to cover every tag whose application should cause a vault re-sync. Candidates to consider:

Tags that affect entry-path in the aggregator (study-group cohorts, practice-community, teacher-training variants)
Tags you set during CRM cleanup / categorization work
Lead-Background, Movement-Industry, Lead-Source tag families

You probably know the full landscape better than the aggregator does — trust your judgement, err on the side of more tags, since each trigger is cheap (the receiver coalesces bursts).

Where to find it

crm.baseworks.com/wp-admin/admin.php?page=fluentcrm-admin#/funnels/42/edit → click the trigger block at the top → add tags → save.

Also worth considering

The automation only triggers on tag applications. If you want custom-field updates (e.g. changes to romaji_name, yjat, pb_primer_deadline) to also fire the webhook, add a second trigger of type “Contact Custom Data Updated” pointing to the field you care about. Each field needs its own trigger row; not a single one covering all fields.

Background

Full setup doc: people-webhook-setup in 03-resources/vault-and-tooling/.

[ ] 2026-04-18 19:45 ET | New FluentCRM custom field: “Romaji name” — populated for 65 contacts

Added by Claude Code on Patrick’s Mac

What’s new

A new contact custom field has been added to FluentCRM on crm.baseworks.com:

Label: Romaji name
Slug: romaji_name
Type: text

It has been pre-populated for every subscriber whose first_name or last_name contains non-ASCII characters — 65 contacts total:

9 diacritic-bearing Western names (e.g. Noémie Normandin → Noemie Normandin, Véronique Pelletier → Veronique Pelletier, Zulma Gámez → Zulma Gamez). Simple NFD strip throughout — no German oe/ae/ue conversion applied.
56 Japanese-named contacts — Hepburn romanization without macrons, Western order (Given Family). Each romanization was reviewed and approved by Patrick against the kanji + any email-address hints. Where the CRM had family-name-in-first-name-field (e.g. id 291 福田/桃子, id 332 品川/隼人), the romaji field rebuilds the Western-convention order.

Why this exists

The vault’s people aggregator builds file slugs from each person’s display name. Japanese (and some diacritic-heavy) names produce unusable slugs — the current MOC at 02-areas/people/index.md has five broken-looking wikilinks like [奈美熊谷](/unknown/) as a result. With the romaji field populated in CRM, the aggregator can pull it via the existing CRM-tag reader in scripts/people_remote.py and use it to generate clean ASCII slugs (nami-kumagai.md instead of a “unknown” fallback). It also gives you a searchable ASCII handle inside FluentCRM’s contact filters.

What you can do with it

Filter/search subscribers by romaji in FluentCRM admin.
Use the field as a merge tag in email campaigns if useful (e.g. for romaji-friendly greetings in bilingual newsletters).
Correct any romanization you disagree with directly in the contact’s profile — the aggregator will pick up changes on its next cron run.

Infrastructure notes

Pre-change CRM DB dump (29 MB) is safely in B2 at b2://cbcrmbaseworks/crm-baseworks-web/adhoc-backups/crm-pre-romaji-field-20260418-193559.sql.
No schema/DDL changes were needed. The field definition is a row in 3ZGRM_fc_meta; values are rows in 3ZGRM_fc_subscriber_meta with key='romaji_name'. Fully reversible with a single DELETE if you ever want to undo.
Changelog entry: ~/Documents/baseworks-changelog/CHANGELOG.md 2026-04-18.

Next step (optional, not done yet)

Wiring build-people-index.py to consume romaji_name for slug generation is the natural follow-up. Let Patrick know if you’d like that done now, or leave for a future session.

[ ] 2026-04-18 18:25 ET | URGENT (only if you pulled between 18:02 and 18:20 ET): changelog repo history was rewritten

Added by Claude Code on Patrick’s Mac

What happened

A commit I made at 18:02 ET (1036717 Step 1: Primer Form 66/67 export plugin) accidentally included ~4,000 files of node_modules/ content from sites/baseworks.com/mockups/ that should have been gitignored. To keep the repo lean, I force-pushed a clean replacement commit (bb65a15) at 18:20 ET. The intended content (a new mu-plugin for Primer forms + SSH docs updates + tightened .gitignore) is in the clean commit.

Do you need to do anything?

Only if you did git pull in the ~/Documents/baseworks-changelog/ repo between roughly 18:02 and 18:20 ET on 2026-04-18. Outside that window there’s nothing to do — your next git pull will just fetch the clean history.

If you’re not sure, run this in Claude Code (or paste it to ChatGPT/Claude Desktop with the directory open):

cd ~/Documents/baseworks-changelog && git log --oneline -5

If you see 1036717 Step 1: Primer Form 66/67 export plugin anywhere in that list, you pulled the bad commit. Fix it by running:

cd ~/Documents/baseworks-changelog
git fetch origin
git reset --hard origin/main

That discards the local bad commit and resets to the cleaned-up remote. No data loss — you had no uncommitted work in this repo, right? If you did have uncommitted changes, stash first (git stash) then reset, then git stash pop.

If 1036717 is NOT in your log, do nothing — you’re already clean.

Why you’re being told

Per our Git Safety Protocol, force-pushing to main is a thing we don’t do casually. I checked the VPS (clean, many commits behind, no problem there) but I can’t verify your machines from here, so this note is the belt-and-suspenders check. Sorry for the jitter.

[ ] 2026-04-18 17:45 ET | People aggregator Step 2.5 complete — CRM tags now flow in automatically

Added by Claude Code on Patrick’s Mac

What landed

The people-profiles aggregator (scripts/build-people-index.py) now pulls remote data on every run and folds it into 02-areas/people/. Your CRM tagging pass is wired in as one of the inputs.

Admin exclusion: Patrick, you, and vendor accounts (brettforge, woologin, etc.) are no longer generated as person files. Configured in 02-areas/people/_admin-exclude.yml.
wp_users join (both sites): emails, display names, user logins, and registration dates are filled in on all 17 person files that had a practice or baseworks user_id. First-name-only records (clementine, dawson, magali, mimi, elinor) collapsed into their full-name counterparts automatically.
wp_bw_activity tier: 19 files now classified active, 47 still unknown (mostly Open Day attendees who never registered — those become inquiry once the contact-form source is wired in).
Fluent CRM reader: pulls tags + custom fields from crm.baseworks.com for 500 tagged subscribers. The mapping file 02-areas/people/_crm-tag-mapping.yml translates your tag titles to entry-path values. Your Tokyo Studio Alumni (tag 103) mapping and YJAT=Yes → teacher-training-graduate custom-field rule are seeded.
BuddyBoss groups: current-programs now reflects Primer Community, Practice Community, and both Montreal Study Group cohorts.

What you can do

Keep tagging in Fluent CRM as normal. Tags flow into person files on the next run (cache TTL is 1 hour; --force-remote refreshes on demand).
If you want new tag titles to drive entry-path, add them to _crm-tag-mapping.yml using the tag title as the key. Example:
```
tag_to_entry_path:
  "Tokyo Studio Alumni": tokyo-studio-alumni
  "TT Graduate 2019": teacher-training-graduate
```
The CRM has 85 subscribers tagged “Tokyo Studio Alumni” and 22 with YJAT=Yes. None of them currently overlap with vault-side person files because Tokyo alumni haven’t engaged through forum/DMs — they’ll get person files once the Step 5 TT-graduate batch import runs, or if any of them interact through the practice site.

What’s next (no action needed from you)

Step 3 is the cron integration so this runs automatically after the forum sync.
Step 1 is the Formidable Primer-segment exports (Patrick owes form IDs).
Step 5 is the TT-graduate batch import.

Full plan: people-profiles-aggregation-plan — Step 2.5 section now marked complete.

[ ] 2026-04-17 11:00 ET | WPRUS admin account overwrite incident — security fix needed

Added by Claude Code on Asia’s Mac Mini

What happened

Today you registered a test account asia@cevaco.co on baseworks.com. WPRUS matched the username asia on baseworks.com to the asia admin account on practice.baseworks.com and synced the email over, locking you out of practice. The plugin was temporarily deactivated on both sites, the email was restored via WP-CLI, the stray user (ID 564, role customer) was deleted from baseworks.com, and the plugin was re-enabled. Access is restored.

Root cause

WPRUS matches users by user_login (username), not email. The asia username exists on both sites. With no role filters configured, any public user who registers on either site with a username matching an admin on the other site can trigger an overwrite of that admin’s data.

Fix needed — two-part

See Patrick’s inbox for discussion. Decisions needed before implementing:

Role filtering in WPRUS — set outgoing_roles and incoming_roles on both sites to only sync subscriber/customer. This prevents any admin account from being touched by the sync, regardless of username collisions.
Rename the asia admin username on practice from asia to pandasia (matching baseworks.com). This eliminates the username collision entirely and is the cleaner long-term fix.

Prior history

This is a pattern — see baseworks-changelog/docs/STAGING-USER-SYNC-ISSUE.md (Jan 2026) and sites/practice.baseworks.com/staging-setup-checklist.md. The staging checklist already deactivates WPRUS after every staging push for exactly this reason. The live site pair (baseworks.com ↔ practice.baseworks.com) has no equivalent protection yet.

[ ] 2026-04-17 09:49 ET | Questionnaire export plugins now emit `user_login` — plus new code-location rules

Added by Claude Code on Patrick’s Mac

Heads-up so you know what changed in the changelog repo and vault today.

What was done

Added user_id, user_login, and profile_url to questionnaire exports on both sites:
- sites/baseworks.com/mu-plugins/bw-questionnaire-export-cli.php (Form 40)
- sites/practice.baseworks.com/mu-plugins/bw-questionnaire-export-cli.php (Form 74)
- Logic: look up the Formidable entry’s user_id, fall back to email if empty. When a WordPress user is found, three new frontmatter fields get written into the 02-areas/method-admin/audience/event-participants/*.md files. Absent users → fields stay absent and the skill falls back to first name.
- Purpose: so the /session-summary skill can rewrite participant first names as BuddyBoss @mentions (e.g. @caitlin_b) in group posts, making forum links click-through instead of plain text. Patrick’s inbox item from 2026-04-13 has the full rationale.
Mirrored the practice plugin into the repo. It had been living only on the server (/var/www/practice.baseworks.com/wp-content/mu-plugins/) with no local copy. It’s now committed at sites/practice.baseworks.com/mu-plugins/bw-questionnaire-export-cli.php. Deployed the modified version back to the server with sudo cp; md5 matches.
Added two new code-location rules to prevent this kind of drift going forward:
- CLAUDE-INSTRUCTIONS.md in baseworks-changelog now has a §3 Mirror-From-Server Checklist: whenever we edit a plugin/snippet/config directly on a server, the final file must be scp-ed back to this repo and committed.
- The vault’s CLAUDE.md now has a new “Code Belongs in the Changelog Repo” section saying site code must live in baseworks-changelog, not here. Vault-native scripts (sync, wikilink check, agent system) are still fine where they are.

Backups

DB backup of practice.baseworks.com taken before deploy: practice-pre-user-login-plugin-20260417-094631.sql (538 MB), uploaded to B2 bucket cbbpracticesite under practice-baseworks-web/adhoc-backups/.
Plugin file backup on server: /tmp/bw-questionnaire-export-cli-practice-backup-20260417-094639.php.

Action for you

Nothing urgent. If you pick up work on the /session-summary skill next, Patrick’s inbox item explains the Pre-Flight map lookup and Phase 4 mention rewrite — the data is now in place for that. If you do server-side plugin edits on any site, please follow the new mirror-from-server checklist.

Commits (main, `baseworks-changelog`)

c6f04c0 Add user_login/user_id/profile_url to Form 40 questionnaire export
e3e89b9 Mirror bw-questionnaire-export-cli.php baseline from practice.baseworks.com
cf2c895 Apply user_login additive change to Form 74 plugin + doc rule

2026-04-16

[ ] 10:15 ET | People Profiles (Phase 2) — full plan drafted, ready for your review

Added by Claude Code on Patrick’s Mac

The Phase 2 plan for the people-profiles aggregation layer has been drafted at people-profiles-aggregation-plan. This builds on the Phase 1 community-forums consolidation (completed 2026-04-14) and creates a person-scoped layer in the vault — one file per person linking to their DMs, forum posts, questionnaire responses, session appearances, and event participation.

Key decisions made with Patrick today:

One-way sync only (CRM/platform → vault, never pushback)
All data committed to git (matching current practice; no new gitignored tier)
Auto-grow with tiers: every unique email gets a person file, labeled active/past/inquiry/mention
Entry-path field tracks how someone first connected (Tokyo studio, TT graduate, Primer, study group, etc.)
Registration-date cutoff for Tokyo alumni heuristic: pre-2024-11-01 (see the registration timeline inbox item below for the data)
Historical backfill: aggregate everything already in the vault (DMs back to 2021); TT graduates without platform accounts are a separate batch import

What we’d like from you: Read the plan and flag anything that doesn’t match your understanding — especially the entry-path categories, the tier definitions, and the privacy approach. No implementation has started yet.

[ ] 09:30 ET | People Profiles (Phase 2) — registration timeline for your review

Added by Claude Code on Patrick’s Mac

Context: Phase 2 planning is underway for the people-profiles aggregation layer. One key design decision is the entry-path field on each person file — how someone first connected with Baseworks. To set the automated cutoff between “pre-existing network” and “current enrollment pipeline,” we pulled a registration-date histogram from practice.baseworks.com. Patrick wants you to see this breakdown and flag anything that looks wrong.

Registration timeline (practice.baseworks.com user_registered):

2016-03 to 2018-09:    8 users   — platform inception / testing
2020-02 to 2020-10:    5 users   — pandemic era
2021-03 to 2021-12:   22 users   — biggest early wave (likely TT + Tokyo alumni onboarding)
2022-02 to 2022-11:    9 users   — trickle (existing network)
2023-02 to 2023-11:    9 users   — trickle
2024-02 to 2024-07:   10 users   — gradual pickup
2024-10:                2 users
2024-11:              42 users   — SPIKE (Open Day / Primer launch)
2024-12:              20 users
2025-01 to 2025-12:   26 users   — steady enrollment
2026-01 to 2026-04:   58 users   — Spring cohort + continued growth
                     ___________
                     ~221 total users

The pattern — two distinct eras:

Pre-November 2024 (~65 users): Low and steady, never more than 8 registrations in a month. These are the pre-existing network — Tokyo studio alumni, teacher-training graduates, early testers, people who found Baseworks through direct outreach or word of mouth.
November 2024 onward (~156 users): Driven by the Open Day / Primer / Study Group enrollment pipeline. 42 registrations in November 2024 alone marked the inflection point.

Proposed automated cutoff: user_registered before 2024-11-01 defaults to tokyo-studio-alumni entry-path (unless a teacher-training graduate list or manual override says otherwise). Users from 2024-11-01 forward fall through to group membership, Primer enrollment, or CRM tags for entry-path derivation.

What we’d like from you: Does this breakdown match your understanding of the user base history? Are there any people in the pre-November-2024 group who you know came through a different pathway (not Tokyo studio or teacher training)? Any corrections or additions to the entry-path categories?

Full plan: [people-profiles-aggregation-plan](/resources/executed-plans/people-profiles-aggregation-plan/) (being drafted now, will be committed shortly).

2026-04-14

[ ] 12:45 ET | Your SSH access to baseworks-n8n was restored + vault capture pipeline fixed

Added by Claude Code on Patrick’s Mac

Why you’re getting this: between 2026-04-05 and today, your SSH keys (both Mac Mini and MacBook Air) were silently missing from ~patrick/.ssh/authorized_keys on baseworks-n8n. If you had tried to ssh baseworks-n8n during that window it would have failed. Everything is restored now — you should have no issues SSH-ing to either VPS from either of your Macs. You don’t need to do anything unless you notice problems.

What happened (short version): Patrick’s morning Slack post to #vault-inbox didn’t land this morning. Investigation found that on 2026-04-05 at 03:23:52 UTC, some unidentified provisioning-style operation wiped ~patrick/.ssh/authorized_keys on both VPSes simultaneously, dropping the machine-to-machine keys that the vault capture pipeline and the n8n daily backup depend on. Your two keys on the n8n VPS were also collateral damage in the same event. The n8n daily backup had been failing silently for 10 days, and the daily infra report had been flagging false-positive “Claude CLI authentication failed” warnings for 4 weeks that masked what was actually broken.

What was fixed today:

All missing SSH keys restored on both VPSes (including yours on n8n)
n8n daily backup ran manually — first successful in 10 days
n8n workflow patched to skip Slack message edits (they were producing empty “File: “ / Tags: / From: unknown” vault capture confirmations)
Claude OAuth token moved to a single source-of-truth file at ~/.config/baseworks/claude-token on baseworks-agents — the 4-week-old false-positive auth warning is gone
A new daily authorized_keys integrity check was added to daily-infra-updates.sh (Test 5) — if any key goes missing again on either VPS, it will show up in both Patrick’s inbox report and #agent-alerts in Slack within 24 hours instead of silently breaking for 9+ days

Where the full record lives:

Shared context doc: claude-code-shared-context.md — full write-up under “Recent Infrastructure Changes → 2026-04-14”, plus a new “SSH Key Canonical Lists” section with fingerprint tables for both VPSes and instructions for adding/rotating machines
Site changelog: ~/Documents/baseworks-changelog/CHANGELOG.md (repo p-oancia/baseworks-changelog, commit b2391fa) — the long-form narrative with root causes, forensic dead-ends, and end-to-end verification

Mystery that remains: whatever wiped the keys on 2026-04-05 is still unidentified. No trace in auth.log (rotated), systemd journal (volatile), bash history, cron, or reboot records. Signature matches a provisioning-style script run from a control node. If you remember doing anything on your Mac Mini or MBA around Saturday evening 2026-04-04 that might have touched user accounts or SSH configs on both VPSes, please mention it — but no worries if not, the new integrity monitor will catch it within 24h if it ever happens again.

No action needed from you unless something doesn’t work.

2026-04-12

[ ] 22:00 ET | Primer Community session — forum reply drafted, tag shortlist reposted, new folder convention

Added by Claude Code on Patrick’s Mac

What: Patrick and Claude worked through several connected items in the Primer Community forum context tonight. Three things worth your awareness: a drafted reply to Nathalie Dore, a repost of the Forum Tag Shortlist to the Primer Community forum, and a new vault organization pattern for hand-curated Primer Community content.

Forum reply drafted for Nathalie Dore. Nathalie posted a question today on the Primer Community forum about keeping the back foot in high-heel position during Applied Practice Lab — Ascendant Torsion, Lesson 3.5. Patrick dictated the substance of the response and iterated several rounds with Claude on tone, structure, and redundancy reduction. The final version addresses her balance/pelvis question directly, then broadens into how the program is meant to be reapplied across labs and why moderation is central. Patrick will post the reply to BuddyBoss when he next sits down; the draft lives in the sidecar (see below) since the synced topic file will be regenerated by the forum content sync as soon as the reply hits BuddyBoss.

Forum Tag Shortlist reposted to Primer Community. The original Forum Tag Shortlist lived only in the Montreal Winter 2026 cohort forum (topic 21433), which meant practitioners in the Primer Community didn’t have direct access to it. Patrick posted a new Primer Community version tonight at: https://practice.baseworks.com/groups/primer-community/forum/discussion/forum-tag-shortlist-how-to-tag-your-forum-posts/ The new version uses grouped bullet clusters (Forms & Practice Structure, Principles & Patterns, Segments S1–S10, Awareness & Perception, Neuroscience & Physiology, Learning & Adaptation) rather than the flat double-spaced list from the original. 61 tags become much more scannable when grouped by category.

New primer-community/ folder convention in the vault. Previously, anything Primer Community-related lived inside the auto-synced subfolders (topics/, groups/, forums/), which meant hand-curated content had nowhere safe to live — the sync script would regenerate those files on every cycle. A new folder at 02-areas/practice-platform/community-forums-groups/primer-community/ now holds hand-curated material specific to the Primer Community forum: draft posts, reference material, curated follow-ups to specific discussions. The folder has its own index.md and is linked from the parent community-forums-groups.md index. The sync script only touches topics/, forums/, groups/, activity/, and direct-messages/, so content in primer-community/ is safe from regeneration.

Curated-notes sidecar pattern. A related discovery: the forum content sync regenerates the entire topic file body on each cycle (not just frontmatter), so any persistent working notes attached to a specific topic need to live outside topics/. A new sidecar pattern lives at primer-community/260412-keeping-back-foot-in-high-heel-position-notes.md, which holds the recommended tag list, the full draft reply as a persistent backup, and working notes for Nathalie’s topic. This sidecar pattern could become standard for any future topic where you or Patrick draft a substantive response: create a matching sidecar in primer-community/ (or a sibling folder for other groups) so the reply and tag choices survive the sync.

Awareness items (no action needed from you):

Sync script gap noted. The forum-content-sync.py script currently doesn’t carry BuddyBoss-level topic tags into the vault frontmatter. Every synced topic ends up with only the static [forum-topic, community, auto-synced] tag set. This is a known gap worth flagging for a future enhancement. For now, the sidecar pattern is our workaround for “recommended tags for this topic” tracking.
CSS styling task deferred. Patrick has a pending task to restyle the BuddyBoss forum tag display on practice.baseworks.com to improve visibility. This is a code-level change to the practice platform and will be handled through the baseworks-changelog repo from the VPS in a future session, not from the vault. No vault notes on this — flagging here only so you’re aware it’s on the pending list.

Files touched in the vault this session:

02-areas/practice-platform/community-forums-groups/topics/260412-keeping-back-foot-in-high-heel-position.md — draft reply added in the Replies section (will be overwritten by sync once Patrick posts on BuddyBoss, which is expected)
02-areas/practice-platform/community-forums-groups/primer-community/ — new folder created
primer-community/index.md — new index for the folder
primer-community/forum-tag-shortlist-repost.md — clean markdown draft used as the source for Patrick’s BuddyBoss repost
primer-community/260412-keeping-back-foot-in-high-heel-position-notes.md — curated-notes sidecar for Nathalie’s topic (holds tag list and persistent draft reply)
02-areas/practice-platform/community-forums-groups/community-forums-groups.md — parent index updated to list the new primer-community/ folder

Next steps for Patrick:

Post the drafted reply to Nathalie’s topic on BuddyBoss and apply the recommended tags from the sidecar
Handle the forum-tag CSS restyle via baseworks-changelog repo from the VPS in a future session
Consider whether forum-content-sync.py should be enhanced to pull topic-level tags from the BuddyBoss API — would remove the need for sidecar tag tracking

Next steps for you (Asia): none immediately. This is an awareness note so you know about the new primer-community/ folder and the sidecar pattern next time you’re working on a forum response. If you find the pattern useful (or not), let us know.

2026-04-08

[ ] 22:15 ET | Forum content sync — major fixes + new features (DMs, translations, thread summaries)

Added by Claude Code on Patrick’s Mac

The forum content ingestion system (System 3, deployed yesterday) had a significant gap: the BuddyBoss REST API doesn’t return group feed posts for private groups. Only 7 of 49 activity posts were being captured. This has been fixed, along with several new features.

What changed:

Activity sync fixed — switched from REST API to direct SQL queries. Now captures all 49 group feed posts + 41 activity comments (previously missing entirely). Activity comments appear inline under their parent post.
Admin DMs now synced — all DM threads involving you or Patrick are pulled into the vault at community-forums-groups/direct-messages/. 42 threads captured. Broadcast messages to cohorts are included. Participant-to-participant private messages are NOT captured.
Automatic English translations — non-English messages (French, Japanese) now show the original text first, followed by an English translation in a blockquote. Runs automatically after each sync.
Thread summaries — settled threads (48h inactive, 1+ comments) get AI-generated structured metadata: topic, category, participants, resolution status, and semantic tags. Runs daily at 1 AM ET.
Permalinks — each activity post now links directly to practice.baseworks.com/news-feed/p/{id}/ instead of showing a raw HTML comment.
Sync frequency — increased from every 2 hours to every 15 minutes.

Vault locations:

Activity digests: community-forums-groups/activity/
DM threads: community-forums-groups/direct-messages/
Full plan: Forum Content Ingestion Plan

Action needed: Review at your convenience. The DMs contain personal participant communications — worth being aware they’re now in the vault (which is private/internal). If you’d like any threads excluded or the admin user filter adjusted, let Patrick know.

2026-04-07

[ ] 22:51 ET | Vault health audit — 149 broken wikilinks + 244 backtick path warnings — cleanup needed

Added by Claude Code on Patrick’s Mac

The new daily vault audit (now reporting to #agent-alerts in Slack every morning at 4:15 AM) surfaced pre-existing broken wikilinks. None are caused by recent work — they’ve accumulated over time. Total: 149 broken wikilinks + 244 backtick path warnings out of 2,492 total links checked.

Broken wikilinks (149 total)

Top missing targets — these account for most of the breakage:

Missing target	Count	Likely cause
`Micro-Movements`	15	File renamed or never created — referenced across multiple vault captures and forum content
`montreal-film-schools-master-list`	13	All in `02-areas/collaborations/photography-videography-2026/contacts/` — planned file never created
`Sense Control Adapt`	11	Baseworks concept — file may have been renamed or needs to be created
`groups/`	9	Partial path references that don’t resolve
`_index`	8	Old-style index references from before the rename to `index.md`
`posting-job-board`	5	Planned file in photography contacts, never created
`study-group-participant-interview-request`	4	Reference to a planned document
`wikilinks` / `wikilink` / `wiki-links`	6	Example/placeholder text in documentation files (KB-Site logs, etc.) — not real broken links

Other categories:

settings-local-json-backup-patrick-mac-mini-2026-04-07 (3) — the JSON backup file exists but isn’t an .md file, so Obsidian can’t resolve it as a wikilink. Harmless.
image.png, filename, Page (6) — example placeholders in documentation, not real links
Various one-off references to files that were moved or renamed during today’s reorganization

Files with the most broken links: contacts-directory.md (photography contacts), additional-resources.md, various campaign files, event-post-system.md, _session-summary-guidelines.md, _campaign-creation-guidelines.md

Backtick path warnings (244 total)

These are paths written as `02-areas/path/to/file.md` instead of [wikilinks](/wikilinks/). Many are in CLAUDE.md, setup guides, voice guides, and technical documentation where backtick paths are appropriate (filesystem references for Claude Code, not navigational links for Obsidian). The audit script flags them but most are intentional per the vault’s own wikilink standards (CLAUDE.md says backtick paths are fine in technical docs).

Recommended approach

Create Micro-Movements.md and Sense Control Adapt.md as Baseworks concept files (or redirect to existing files if they were renamed) — you’d know best whether these should be standalone concept files or aliases to existing content
Create montreal-film-schools-master-list.md or remove the dead links from the photography contacts directory
Fix the _index → index references (8 leftover from the earlier index file rename plan)
Ignore placeholder links in documentation files (wikilinks, image.png, filename, Page) — these are examples in KB-Site logs and implementation docs
Run python3 scripts/check-wikilinks.py after fixes to verify the count drops

To check the full list at any time: python3 scripts/check-wikilinks.py To check only recently changed files: python3 scripts/check-wikilinks.py --changed

[ ] 22:26 ET | Claude Code settings cleanup — please run on your machines

Added by Claude Code on Patrick’s Mac

Patrick’s .claude/settings.local.json on his Mac Mini was cleaned up — 52 accumulated permission entries reduced to 21. Most removed entries were one-time session artifacts, redundant patterns, and WebFetch domains for tools we researched but never adopted.

Action needed: When you have a moment, run the same cleanup on your MacBook and Mac Mini. Start a session in the vault and ask Claude Code to review .claude/settings.local.json and streamline it using the same approach documented here:

Full plan and rationale: 03-resources/executed-plans/settings-local-json-cleanup.md
Patrick’s Mac Mini backup (for reference): 03-resources/executed-plans/settings-local-json-backup-patrick-mac-mini-2026-04-07.json

Patrick’s MacBook Pro also needs the same cleanup (it wasn’t open during this session).

The VPS doesn’t have a settings.local.json (it uses dangerousMode), so no cleanup needed there.

This is low priority — all machines work fine as-is. The cleanup just removes clutter from old sessions.

[ ] 21:35 ET | Resources reorganized + upcoming consolidation audit — your input welcome

Added by Claude Code on Patrick’s Mac

Two things here: a completed reorganization, and an upcoming piece of work where your input would be valuable.

1. `03-resources/` is now organized

The 14 loose files that were sitting directly in 03-resources/ have been sorted into topical subfolders:

vault-and-tooling/ — media storage plan, media organization memo, Google Workspace CLI setup, sitemap reference
translation/ — all three WPML plans + Japanese translation conventions
community/ — forum moderation plan (still awaiting your review from the earlier inbox item)
research/ — Karpathy LLM wiki pattern evaluation
executed-plans/ — completed plans preserved for reference (vault SQLite sync, forum ingestion, index rename fix, permission pattern fix)

Each subfolder has an index file (00-{folder}-index.md) that sorts to the top and shows a descriptive name in Obsidian’s graph view. All wikilinks, skill file paths, and script paths have been updated. Nothing should be broken.

2. Upcoming: instruction consolidation audit — your input welcome

During the reorganization, Patrick and Claude identified that instructions, operational references, and folder indexes overlap across multiple locations in the vault. Examples:

Voice guide loading instructions appear in both CLAUDE.md files AND voice-guides/README.md
Server/infrastructure details appear in both agent-system/README.md AND 00-inbox/claude-code-shared-context.md
The claude-code-skills/index.md in Obsidian may be redundant with the actual .claude/skills/ folder
Some README files are instruction guides (not indexes) and might belong as skills or in CLAUDE.md instead
19 index.md files across 02-areas/ have generic names that are ambiguous in graph view and fragile for wikilink resolution

The next session will audit all of this and propose a consolidation plan. Before that happens, if you have thoughts on any of the following, add them here or flag in Slack:

Do you reference the voice-guides/README.md directly, or do you just load the guides via CLAUDE.md?
Do you ever open the claude-code-skills/index.md in Obsidian to find skills, or do you use / commands?
Any folders where you find the naming confusing or hard to navigate?

No urgency — this is informational. Patrick has the same item in his inbox.

2026-04-06

[ ] 17:00 ET | Google Workspace CLI setup — persistent auth ready for you

Added by Claude Code on Patrick’s Mac

Patrick set up the Google Workspace CLI (gws) with persistent authentication for the baseworks.com domain. The RAPT (re-authentication) policy has been changed to “never expire,” which means once you log in, you stay logged in indefinitely. No more repeated browser auth flows.

What you need to do

1. Install gws (if not already installed):

brew install gws

2. Authenticate once:

gws auth login

This will open a browser window. Log in with asia@baseworks.com and grant the requested permissions. After that, you should never need to do this again.

3. Test it:

gws gmail users messages list --params '{"userId": "me", "maxResults": 1}'

If you get a JSON response with a message ID, you’re good.

What this gives you

gws gmail — read, send, draft emails
gws drive — list, read, manage files and shared folders
gws calendar — manage calendars and events
gws docs — read and write Google Docs
gws sheets — read and write spreadsheets
gws slides — read and write presentations
gws tasks — manage task lists

Claude Code can use all of these via the gws command. This replaces the need for the Google Workspace MCP connection.

Security notes

Credentials are stored encrypted in your macOS Keychain (not as a file on disk)
Protected by FileVault and your login password
The “never expire” RAPT setting was applied domain-wide by Patrick in the Google Admin Console — it applies to both accounts automatically
If you ever need to revoke access: gws auth logout

If you have issues

If you get an invalid_grant error, just run gws auth login again. With the new RAPT setting, this should be a one-time event rather than a recurring annoyance.

[ ] 11:00 ET | Content repurposing and memes — your input needed

Added by Claude Code on Patrick’s Mac

Patrick is working on building out campaign archives and content strategy. During this session we documented the January 16, 2026 newsletter (your “What Movement Training Misses” article) as a full campaign in the vault, and started a framework for a new blog post by Patrick expanding the Brain Fodder on learning styles.

This led to a broader conversation about content repurposing across social media — specifically the idea of creating reusable “meme” units from our published content. Patrick wants your input before we develop this further.

The meme concept

Not traditional internet memes. These would be reusable conceptual units — a single idea, stat, or contrast extracted from blog posts and newsletters that can be rendered across formats:

Quote cards / carousel slides for Instagram
Short caption posts for Facebook
Pull-quotes for newsletters
Standalone social media posts that can rotate through feeds over time

Examples from your article alone

Your “What Movement Training Misses” article contains several natural meme units:

“97-99% of movement instruction focuses on what you can see” — the core stat that frames the gap
“66% still experience chronic tension despite 5+ modalities” — the participant data point
“You can’t watch your way through distributed activation” — the simultaneity argument
The Baseworks vs. yoga vs. Feldenkrais comparison (60% vs. 1-3% vs. 27%) — could work as a simple chart or infographic card

Why this matters

Social media feeds are ephemeral — content disappears within hours. Repurposing lets us re-enter the same ideas from different angles over time without repeating ourselves. The strategy aligns with our niche-building approach: informational, substantive content that the right people recognize and engage with. Not promotional, not sensational — just the ideas themselves, presented clearly.

What we need from you

Quality control: These units would draw from your scientific framing and data. You should have a say in which claims are suitable for standalone presentation (out of full article context) and which need their surrounding context to be accurate.
Tone check: Making sure compressed versions of your arguments don’t lose precision or drift into oversimplification.
Ideas: Are there other reusable units from your existing or planned writing that you think would work well in this format?

Patrick said he’d like to discuss this with you before we build anything. No action needed now — just flagging for when you have time to think about it.

Campaign: What Movement Training Misses
Blog: What Movement Training Misses
Blog framework: Learning Styles Gap (Patrick’s new article)
_campaign-creation-guidelines — current campaign structure (no meme/card system yet)

2026-04-01

[ ] 22:55 ET | Questionnaire system deployed to practice platform — Phase 2 plan

Added by Claude Code on Patrick’s Mac

Patrick and Claude deployed the Event Participation Questionnaire system to practice.baseworks.com tonight. Summary of what’s live:

Background profile tab — new “Background” tab on BuddyBoss profiles with the questionnaire form (same form as on baseworks.com). URL: /members/me/background/
Persistent prompt banner — bottom-right notification card for enrolled participants who haven’t completed the form. Tag-based (Study Group + Practice Sessions tags). X hides for current page; “Later” dismisses for 24 hours.
Dashboard teaser — [bw_dashboard_teaser] shortcode shows muted preview of Primer + Activity dashboard for practice-session-only users with CTA linking to baseworks.com/programs/
Obsidian export — WP-CLI command on baseworks.com exports Form 40 entries to event-participant markdown files in the vault. Running 6x daily via cron on agents VPS.
Staging restored — pracstage.baseworks.com pushed from production, post-push cleanup done, fonts fixed.

Phase 2 items (not yet started):

Groups teaser page — muted group cards for practice-session-only users (Elementor page + shortcode + tag-based menu visibility)
Soft-gate form completion — require questionnaire before certain course/community milestones
Real-time Obsidian sync via webhook (replace cron)
Custom multi-step form replacing Formidable
Broader data accumulation from forum/DM/email

Files (in baseworks-changelog repo):

sites/practice.baseworks.com/mu-plugins/bw-profile-background-tab.php
sites/practice.baseworks.com/mu-plugins/bw-questionnaire-prompt.php
sites/practice.baseworks.com/mu-plugins/bw-dashboard-teaser.php
sites/baseworks.com/mu-plugins/bw-questionnaire-export-cli.php

No action needed from you right now — this is informational so you have full context.

[ ] 15:00 ET | Google Calendar setup — Baseworks MTL Sessions + participant-facing calendars

Added by Claude Code on Patrick’s Mac

Patrick set up Google Calendar integration for all Spring/Summer 2026 studio sessions. Three calendars now exist under pat@baseworks.com:

1. Baseworks MTL Sessions (private — you and Patrick)

Combined calendar with all Study Group + Practice Session events
Shared with your Workspace email
Color-coded: teal for Study Group, green for Practice Sessions
Subscribe via iCal URL from the calendar’s Integrate settings (Settings → Integrate calendar → Secret address in iCal format)

2. Baseworks Study Group — Spring 2026 (public — for participants)

7 sessions, April 4 – May 16
Set to public so participants can subscribe without seeing pat@baseworks.com

3. Baseworks Practice Sessions — Spring/Summer 2026 (public — for participants)

13 sessions, April 4 – June 27
Same public setup

All events include:

Session/week number + studio number (Studio 1 or Studio 2) in the title
Full address: Proto Studio, 5333 Av. Casgrain, suite 1107, Montréal, QC H2T 1X3
Google Maps directions link in the notes (workaround for Apple Calendar not making the location field clickable)

Action needed:

Confirm the private calendar (Baseworks MTL Sessions) is visible in your Google Calendar — Patrick shared it with you
Grab the public iCal URLs for the two participant-facing calendars (Settings → Integrate calendar → Public address in iCal format) — these need to be added to the Practice Platform as “Add to Calendar” links for participants
Note on Apple Calendar: The location field is not clickable when synced from Google Calendar to iCal — this is an Apple limitation. The Google Maps link in the notes is the workaround.

2026-03-27

[ ] 22:30 ET | Media CDN pipeline live — WPRUS test needed + new media workflow

Added by Claude Code on Patrick’s Mac

The Baseworks media CDN is now operational at media.baseworks.com. All images are served from Backblaze B2 via Cloudflare CDN. The Obsidian vault will no longer store binary image files — instead, markdown files reference images via CDN URLs (e.g., ![alt text](https://media.baseworks.com/path/to/image.webp)).

Action needed:

Test WPRUS cross-site login sync with swantest2 (swanonthecliff@gmail.com) — Cloudflare proxy was re-enabled for baseworks.com, practice.baseworks.com, and crm.baseworks.com. The /etc/hosts entries on the server are still in place, so WPRUS should still work. Please confirm sync works as expected.
Review the media organization memo at media-organization-memo — this outlines how photos will be organized going forward (storage vs. discovery, naming conventions, manifest system). Your input on the forms reference library and taxonomy is especially valuable since you work closely with the photo assets.

What changed:

/compress-photos skill now handles the full pipeline: resize originals → create WebP + JPEG → upload to B2 → output CDN URLs
All existing vault images (102 files) are now on the CDN
Markdown files reference CDN URLs; images render inline in Obsidian as before
NAS (/volume1/baseworks/media/) is the local workspace for browsing and sorting photos before upload

No action needed on:

The CDN setup itself — it’s working
Existing campaign files — already updated

2026-03-25

[ ] 23:45 ET | Synology SSH — your keys are authorized, ready to test

Added by Claude Code on Patrick’s Mac

Both your SSH public keys have been added to the Synology NAS. ssh synology should now work from both your Mac Mini and your MacBook Air without a password.

Action: Test from each machine:

ssh synology "echo connected"

If either fails, let us know which machine and the error message.

Update (2026/03/25): both failed, Patrick notified.

[ ] 23:15 ET | Winter Study Group Campaign — Final Revision Complete, Ready for Your Review

Added by Claude Code on Patrick’s Mac

The Winter Study Group campaign has been fully revised. All campaign content (social posts, newsletter feature, LinkedIn article) has been updated to match the current blog post (8th draft, 2026-03-25), the latest voice guide (v2.0–2.2), and Patrick’s content strategy direction. The ball is in your court.

What changed in this revision pass

Factual corrections (matching current blog post):

“massage therapist” → “movement educator with a yoga teaching background” across Facebook, Instagram, and vignettes table
LinkedIn article cut point quote updated (“the in-person sessions” not “in-person classes”)
“Six weeks” → “Seven weeks” in Facebook and Instagram
All “practised/practising” → “practiced/practicing” (American English per voice guide v2.1)
Blog draft count updated to 8th draft
Alt text spelling corrected

Blog post:

H2 subtitle/deck added below the H1 for AI discoverability: “What the Winter 2026 cohort in Montreal looked like, what participants reported, and what we observed.”
No other changes to the blog post

Tone and strategy revision (Patrick’s direction):

Newsletter lead-in, LinkedIn feed post, and Facebook opening all rewritten to mirror the blog’s documentary register — contextual framing about the Tokyo studio history and the developmental question (can the method be made learnable without that daily studio backdrop?), not promotional hooks
LinkedIn feed post now frames why the article matters, not just what it covers
Facebook and Instagram marked as collaboration posts between Patrick’s personal account and the Baseworks account, with first-person attribution in the closing (“I wrote about…”)

LinkedIn publishing structure:

Article to be published from Patrick’s personal account (not Baseworks company page) for better reach
Byline: “Patrick Oancia and Ksenia Shcherbakova” in the article body
Canonical URL set to baseworks.com article so search engines credit the original
Baseworks company page reshares

Campaign guidelines updated:

New “Content strategy: niche-building, not mass appeal” section added to _campaign-creation-guidelines.md
Publishing sequence updated with LinkedIn canonical URL, collaboration post format, and personal account publishing as standard practices
Platform tone table updated

What needs your review

Full campaign copy review — Read through all social posts (LinkedIn article + feed post, Facebook, Instagram) and the newsletter feature section. Confirm the tone, voice, and factual accuracy. Files:
- Campaign note: 02-areas/communications/campaigns/2026-03-winter-study-group-campaign/2026-03-winter-study-group-campaign.md
- Newsletter: 02-areas/communications/newsletters/2026-03-newsletter/2026-03-newsletter.md
- Campaign guidelines: 02-areas/communications/campaigns/_campaign-creation-guidelines.md
Create Brain Fodder image variations — The Brain Fodder section (“What are you not noticing?”) needs image assets for the newsletter and social platforms. Place them in: 02-areas/communications/campaigns/2026-03-winter-study-group-campaign/2026-03-winter-study-group-recap-assets/
Practice Platform links decision — The newsletter (Section 4 and footer) links directly to practice.baseworks.com. Per the linking rule, public traffic shouldn’t go there since the app is gated. Should these link to baseworks.com instead, or is it acceptable in a newsletter context? Flag for Patrick if unsure.

[ ] R&D documentation — separate method development from program development

Added by Claude Code on Asia’s Mac Mini

Context: While reviewing the Winter Study Group blog post, Asia noted that the standard R&D sequence description conflates two distinct conceptual tracks that should be documented separately.

The conflation: The sequence (digitize syllabus → build Practice Platform → develop Primer → design hybrid format → first Study Group) is correct chronologically, but mixes:

R&D of the Baseworks Method itself — the movement pedagogy, principles, taxonomy, forms vocabulary (the what)
R&D of the programming/delivery format — Study Groups, Study Labs, hybrid format, Practice Platform as a delivery vehicle (the how we teach it)

What to do: Develop a documentation note (or expand an existing one in 02-areas/method-admin/core/) that clearly separates these two tracks. The goal is that any future description of the R&D phase can point to this distinction — so blog posts, the About page, and program pages don’t accidentally collapse the two.

Coordinate with Patrick on the right location and scope before writing.

[ ] Synology SSH — test `ssh synology` on both machines after Patrick authorizes keys

Added by Claude Code on Asia’s MacBook Air

Patrick’s inbox (2026-03-25) has both public keys and instructions to add them in Synology DSM. Once he confirms:

Mac Mini: run ssh synology in terminal — should connect without a password prompt
MacBook Air: run ssh synology in terminal — should connect without a password prompt

Both machines have Tailscale connected and the ssh synology alias configured.

2026-03-18

[ ] 16:15 ET | New campaign infrastructure + /create-campaign skill — setup instructions for your machines

Added by Claude Code on Patrick’s Mac

What happened: Patrick built a campaign/newsletter content infrastructure in the vault today. There is now a structured workflow for creating content campaigns (blog + newsletter + social media) that are fully wikilinked into the knowledge base. A Claude Code skill (/create-campaign) automates the process.

What’s already done (no action needed from you):

Campaign and newsletter directory structure under 02-areas/communications/
First campaign note (Winter Study Group Recap) and first newsletter issue (March 2026) created
Campaign creation guidelines: _campaign-creation-guidelines
Campaign index (editorial calendar): _campaign-index
The skill file is checked into the vault repo at .claude/skills/create-campaign/SKILL.md — it arrives on all machines via git pull

VPS: Already set up. The skill file was pulled to /srv/baseworks/knowledge-base/ and both patrick and asia users have access. When you SSH into the VPS and start Claude Code in the vault directory, /create-campaign will be available.

What you need to do on your Macs (~5 min per machine):

Do this on Mac Mini, M1 Laptop, and MacBook Air:

Pull the latest vault changes:
Terminal window
```
cd ~/Obsidian/baseworks-kb-shared-brain
git pull
```
This brings in the .claude/skills/create-campaign/SKILL.md file and all the new campaign structure.
Verify the skill is visible:
- Open the vault folder in VS Code
- Start or open Claude Code in the sidebar
- Type / — you should see create-campaign in the autocomplete list
- If it doesn’t appear, reload the VS Code window: Cmd+Shift+P → “Developer: Reload Window”, then try again
Test it (optional):
- Type /create-campaign Test — Claude will start the campaign creation workflow
- You can cancel after confirming it loads — this is just to verify it works
- The skill will walk you through: KB research → planning → file creation → index updates → wikilink validation

How to use it going forward:

When you or Patrick want to create a new campaign: /create-campaign [topic]
Example: /create-campaign Spring 2026 Study Group Enrollment
Claude will research the KB, propose a plan, and walk through the full creation process
The guidelines document has the full reference: _campaign-creation-guidelines

How skills work in Claude Code:

Skills are .md files in .claude/skills/skill-name/SKILL.md inside the project
They show up in the / menu when you start typing
They’re checked into git, so they sync across all machines automatically on git pull
You can create new skills for any repeatable workflow — just create a new folder under .claude/skills/ with a SKILL.md file

Browsing skills inside Obsidian:

The .claude/ folder is invisible to Obsidian (it’s a dotfolder)
All skills are indexed at Claude Code Skills — this is the Obsidian-visible reference
Each skill has a guidelines document somewhere in the vault (linked from the index) — the human-readable version of what the skill does
When you create a new skill, add it to both .claude/skills/ (for Claude Code) and the skills index (for Obsidian visibility)

Requested action: Run git pull and verify the skill appears on at least one of your machines. No rush.

[ ] 15:30 ET | Research needed — Google Drive media linking in Obsidian campaign notes

Added by Claude Code on Patrick’s Mac

Context: Patrick is building a campaign/newsletter infrastructure in the vault under 02-areas/communications/. Campaign notes will reference images via standard markdown image links rather than storing binary files in the git repo. The preferred workflow: find images in Google Drive (via Gemini or the anti-gravity agent manager), then drag or paste the reference into the campaign note in Obsidian so images render inline.

The question: When you drag or paste a Google Drive image reference into an Obsidian note, does it arrive as a renderable image URL or as a sharing page link? Google Drive sharing links (https://drive.google.com/file/d/ID/view) won’t render inline in Obsidian — only direct image URLs do (https://drive.google.com/uc?export=view&id=FILE_ID).

What to test (either machine):

Find an image in Google Drive
Try dragging it / copy-pasting the link into an Obsidian note
Check: does the image render inline, or do you just see a link?
If it doesn’t render, try rewriting the URL as https://drive.google.com/uc?export=view&id=FILE_ID and see if that works

Why this matters: The new campaign structure stores all image references directly in campaign notes (no separate manifest files). If Drive URLs need conversion, we’ll need to either automate that or use an alternative host (the VPS FileBrowser at files.baseworks.com would give direct URLs natively).

Requested action: Test when you have 5 minutes and report findings. No rush — this is a workflow question for the campaign infrastructure, not blocking anything immediately.

2026-03-07

[ ] 12:30 ET | Quebec language compliance — French checkout flow and terms pages needed

Added by Claude Code on Patrick’s Mac

Context: Patrick completed a Quebec language compliance review (Bill 96 / Bill 101). The course content is exempt (adult education), but the commercial materials — terms, privacy policy, enrollment flow — need French versions. This is Phase 1 of a phased plan.

Full compliance roadmap: Quebec Language Compliance — Bill 96

What needs to be implemented (Phase 1):

1. French terms checkbox on all Montreal checkout entry points:

The goal: before any add-to-cart redirect, users see a bilingual checkbox with a link to the French terms. The CTA button stays hidden/disabled until the checkbox is checked.

Checkbox text:

J’ai lu et j’accepte les conditions générales / I have read and accept the Terms and Conditions

Three places this needs to go:

Entry point	Where in code	What to modify
Practice Sessions intake form — Outcome A card	`baseworks-changelog/sites/baseworks.com/code-snippets/2026-03-05-practice-sessions-quiz-v2.php`	Add checkbox + terms link before the “Purchase Intro Session Pass” CTA in the outcome HTML. Hide CTA until checked.
Practice Sessions booking widget — alumni/standard	`baseworks-practice-sessions.php` mu-plugin on baseworks.com	Add same checkbox before the checkout button in the widget output
Study Group — “Join the Program” button	Elementor page for study group event	Add checkbox above the CTA (either Elementor edit or shortcode wrapper)

Technical reference: Practice Sessions Infrastructure — Known Gaps

2. French practice sessions page with French intake form:

Duplicate the practice sessions page as a standalone French WordPress page (not WPML)
Duplicate the [bw_practice_quiz] shortcode with French strings — either [bw_practice_quiz_fr] or a lang="fr" parameter
The French Outcome A card should include the checkbox natively with French terms linked
Cross-link between English and French pages

3. French policy pages (Patrick will handle the translation, you handle the WordPress pages):

5 standalone WordPress pages with translated policy content (not Termageddon widgets, not WPML)
Patrick will provide the translated text; you create the pages and add cross-links between English and French versions

Not needed for Phase 1:

No WPML French language addition
No WooCommerce cart translation
No custom WooCommerce plugin
Professional translation review comes in Phase 2

Requested action: Review this plan and flag any implementation concerns. No rush to start — Patrick is still working through the compliance roadmap and will coordinate timing.

Completed

2026-08-03

2026-08-03 11:20 ET — Shobayashi invoice settled, and a reply went out in your name

2026-08-03

2026-04-23 21:45 ET — Shobayashi invoice № 2026-05190 (BASEWORKS JPO mark 5823947) unpaid — decision needed

2026-07-24

2026-07-16 14:24 ET — BRNet recap Instagram carousel scheduled (collaborator post with BRNet)

2026-07-09 13:03 ET — Review the BRNet recap LinkedIn article before it publishes

2026-07-01 14:51 ET — New blog post draft ready for your read — subtitle input wanted

2026-07-24

2026-07-22 18:56 ET — FYI: email “view in browser” links were broken on mobile — fixed with a temp snippet on crm

2026-07-22 ET — Substack — Patrick’s take (and Claude’s) on whether we should start one

2026-07-14

2026-07-08 12:42 ET — DOCK 11 Berlin: 2027 reschedule in progress, reply to Katja sent

2026-07-07 10:05 ET — Round 15 voice revision on “What’s Between the Notes” — your review requested; voice guide v2.8 draft pending your confirmation

2026-07-03 16:08 ET — Proto Studio fall booking confirmed + residency rate now documented

2026-07-01

2026-07-01 17:20 ET — What’s Between the Notes: your Round 10 comments are all resolved — please review

2026-06-26

2026-06-25 ET — BRNet 2026 photos uploaded — researcher names needed for 6 slides

2026-06-24 16:59 ET — BRNet 2026 article — Round 3 edits complete, ready for your review

2026-06-19

2026-06-18 00:30 ET — BRNet 2026 communications post: Asia completed collaboration pass. Round 2 edits done, Version 1 blocks archived. Handed off to Patrick for venue notes, audio gap (02:26–05:08), → format publishing decision, and final sign-off. 2026-06-brnet

2026-06-17 23:00 ET — BRNet 2026 context doc (work in progress notice) — superseded by 00:30 ET item above.

2026-05-20

2026-05-19 17:30 ET — Session 7 summary: finalized, ready for forum post and group post

2026-05-13 21:30 ET — Meghan Moe Beitiks — send course access denial response

2026-05-20

Noemi Normandin — second system email to send immediately

2026-04-10

New ACF field needed: “Last Updated” date on article post type

2026-04-08

Forum content sync live + moderation plan — your review needed

Vault infrastructure upgrade — what changed and why

Vault cleanup — rogue files removed, folders reorganized

2026-04-07

Vault sync upgrade — one-time setup needed on your machines

2026-04-06

Staging sites causing 502s on production — Automator crons need attention

Session 1 — edits accepted, forum post live

2026-04-05

2026-04-05 12:00 ET — Session 1 summary reviewed and edited. Factual corrections, major additions (Distributed Activation, Gurdjieff, Ignition paragraphs marked in red), editorial summary added at bottom. Session 1 prep response added. Winter 2026 Learnings read and annotated. Feedback delivered to Patrick’s inbox. Session 1 Summary

Session 1 prep notes for tomorrow — please review

Perception Gap blog — LinkedIn campaign published

2026-04-01

Set up B2 authorization for /compress-photos on your Mac Mini

2026-03-25

2026-03-25 — Writing Process + Voice Guide Structural Changes — informational, archived. Archive

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

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

[ ] 2026-07-18 19:22 ET | Article images: always use the -scaled URL, never the raw 4K upload

Added by Claude Code on Patrick’s Mac

Images on “What’s Between the Notes” were intermittently showing broken-image icons on iPhones. Cause: the post embedded the raw 3840px WebP URLs, so WordPress emitted no srcset and phones had to decode ~400MB of 4K bitmaps — over WebKit’s per-tab memory limit. Fixed on the live site (post 61994), and no other posts were affected (audited everything since March).

Two skills were updated — pull the vault before your next publishing session:

publish-article: get image URLs with wp_get_attachment_url() (the old --field=guid step returned the raw original — that was the bug). A post-publish verification step now checks every image has srcset.
compress-photos: new “4K outputs and device memory” section explaining the rule.

Action: read the two updated skill sections before publishing the next article with photos.

[ ] 2026-08-12 15:50 ET | Unified Voice Guide v3.0 draft: two new sections for your confirmation

Added by Claude Code on Hutch Mini

Patrick reviewed the guides against three preferences: no reciprocal praise, no unnecessary self-commentary, sparing use of quotation marks and long dashes. Two of those were already covered, two were not. VOICE-GUIDE-UNIFIED is now at v3.0 draft with two additions, both pending your confirmation per the governance table.

Core Principle 9, “Acknowledge the content, not the person.” Respond to what someone said, not to the person or the quality of their engagement. “That’s striking” and “We’re genuinely glad you reached out” both work; “Great question” and “That’s a rare observation, and it says a lot about how you’re engaging with the practice” do not. This generalizes the warmth-through-specificity rule already in Patrick’s guide and the no-editorializing rule already approved for group posts. The group-post section stays the stricter form. The forum guidance under By Context changed from “validate the contribution first” to “acknowledge the substance of the contribution first.”

6b, “Emphasis: quotation marks and the alternatives.” Quotation marks stay available and are not restricted. The point is density: in program copy, italics, bold, or a link may serve better depending on context; in short correspondence and forum replies, keep them occasional. Patrick’s framing was that em dashes and quotation marks together are too much for human-to-human messages.

Also cleaned: five pieces of model copy in the unified guide still carried em dashes (the two transfer-language blockquotes and three Prefer: lines). Explanatory prose in the guide is unchanged, since these are internal docs.

One item in your guide, for you to apply or decline. VOICE-GUIDE-KSENIA-ASIA line 36 has an em dash inside sample copy: “I saw your note about timing — here’s what I’ve done”. Suggested: “I saw your note about timing. Here’s what I’ve done”. The 2026-07-07 skills audit flagged this and it never landed. Your guide, so it wasn’t touched.

Neither section is settled. Patrick’s view is that reciprocal praise in particular depends on how a sentence is built, which is why Principle 9 draws the line at evaluating the person rather than banning warmth outright, and why 6b sets no limit on quotation marks. Both are expected to be refined once they’ve been used for a while. Push back on the wording wherever it doesn’t hold up in your own writing.

Action: confirm or adjust Principle 9 and 6b, and decide on the line in your own guide.

Asia's Inbox

Open Items

[ ] 2026-08-03 18:18 ET | Vault session records: no changelog file (Patrick’s decision) — check your machines’ global CLAUDE.md

[ ] 2026-07-31 10:05 ET | How our Claude Code setup is organised — and why your global settings must not override it

The principle

⚠️ What this means for your setup

Why we bothered

Where things are

One thing that happens automatically in the site-build repo

On the site build specifically

[ ] 2026-07-30 22:15 ET | Bricks Accelerator course: start weeks 1–3 on your own WordPress install

⚠️ Please stop at the end of week 3

Your own WordPress install — on our server, not InstaWP

Reference material, if you want it

[ ] 2026-07-18 22:52 ET | Mehling guided read: Patrick’s answers through Movement 7 are ready for your reactions (dialogue still in progress)

[ ] 2026-07-18 11:54 ET | New research-network contact: Ruth Knight (BRNet panel speaker)

[ ] 2026-07-16 13:06 ET | New LinkedIn company page Overview + brand guide: review

[ ] 2026-07-08 17:57 ET | Media drop-off: new photos go in media/incoming/, not media/inbox/

[ ] 2026-07-07 11:20 ET | Audit decisions executed: publish gate, PARA restructure, plans moved to 01-projects

[ ] 2026-07-07 15:45 ET | Unified voice guide v2.9 draft — group-post neutral register, pending your confirmation

[ ] 2026-07-07 10:30 ET | Vault audit quick wins done — one small check needed on your machines

[ ] 2026-07-05 11:16 ET | Comprehensive vault audit complete: report ready, notes wanted (no action needed)

[ ] 2026-07-04 22:14 ET | Forest hike conversation: transcript + independent analysis ready

[ ] 2026-06-30 ET | Framework proposal for your reaction: two axes, not two layers

[ ] 2026-06-27 ET | Two new skills added: /happyfiles-folder and /wp-media-import

[ ] 2026-06-24 16:22 ET | Marta Stothers response — revised draft ready for your review and send

[ ] 2026-06-10 09:00 ET | Interdisciplinary conference targets — new reference file

[ ] 2026-06-02 13:40 ET | Pull before you push — Triple Irony campaign note was edited

[ ] 2026-05-19 20:27 ET | Policy draft: Course Participation and Platform Integrity — review needed

[x ] 2026-05-19 20:08 ET | Noemi Normanding — course progress reset, email needs sending

[ ] 2026-05-19 14:35 ET | SSH between your Mac Mini and MacBook Air — one-time setup

[ ] 2026-05-15 14:34 ET | baseworks-new-site-build — one-time machine setup required for /capture-content

[ ] 2026-05-15 ET | Fluent Community — WPManageNinja response received, please review for workarounds

[ ] 2026-05-14 14:56 ET | Fluent Community migration viability — for awareness and input

[ ] 2026-05-14 16:30 ET | post-to-group skill updated — voice compliance check now required

[ ] 2026-05-13 15:30 ET | Press Archive Analysis — complete (all 142 PDFs)

[ ] 2026-05-13 | Press Page Brief — two-layers framework integrated; D13/D14 likely resolved by your framework document

[ ] 2026-05-12 14:45 ET | CLI notification pipeline — fully complete + participant communications workflow direction

Part 1 — Correction to the 2026-05-10 notification item

Part 2 — Participant communications: use the skills, don’t post directly

Part 3 — What was built today (changelog reference)

[ ] 2026-05-10 ET | FYI: Forum reply notification pipeline fixed — what changed for you

[ ] 2026-05-09 18:00 ET | Action needed: install /watch skill on your machines

[ ] 2026-05-07 ET | New global skill available: /watch for video analysis

[ ] 2026-05-05 16:30 ET | baseworks.com registration — bot protection updated, keep an eye on new signups

[ ] 2026-05-05 13:30 ET | Language as a Teaching Medium blog post — defer to new site + consider LinkedIn instead

[ ] 2026-05-04 11:00 ET | Session 5 Summary draft ready for review

Where to find it

What’s in it

What’s needed from you

Related

[ ] 2026-04-22 21:15 ET | Patrick’s new article “Sensitivity as a Trainable Skill” is live — mentions your research and upcoming talks

What’s live

Where you appear

For your review

Related — voice guide v2.7 draft pending governance review

Requested action

[ ] 2026-04-22 15:30 ET | New collaboration protocol for /draft-response

What changed

What this means for you in practice

The protocol doc

Why this matters

Requested action

Related

[ ] 2026-04-18 20:45 ET | Widen trigger-tag list on FluentCRM automation #42

Context

What’s needed from you

Where to find it

Also worth considering

Background

[ ] 2026-04-18 19:45 ET | New FluentCRM custom field: “Romaji name” — populated for 65 contacts

What’s new

Why this exists

What you can do with it

Infrastructure notes

Next step (optional, not done yet)

[ ] 2026-04-18 18:25 ET | URGENT (only if you pulled between 18:02 and 18:20 ET): changelog repo history was rewritten

What happened

Do you need to do anything?

Why you’re being told

[ ] 2026-07-08 17:57 ET | Media drop-off: new photos go in `media/incoming/`, not `media/inbox/`

[ ] 2026-05-07 ET | New global skill available: `/watch` for video analysis

[ ] 2026-04-22 15:30 ET | New collaboration protocol for `/draft-response`

[ ] 2026-04-17 09:49 ET | Questionnaire export plugins now emit `user_login` — plus new code-location rules

Commits (main, `baseworks-changelog`)

1. `03-resources/` is now organized

[ ] Synology SSH — test `ssh synology` on both machines after Patrick authorizes keys