Taxonomy 101 Lesson 2 of 2

The D8–D12 bless walk.

Five decisions SL-5.3 just locked. Each one as a card: the question, what was on the table, what got picked, what it actually means for you. The "your bless" prompt at the bottom is the spot to pause and ask yourself: does this still make sense?

Bless walk · 5 cards · ~6 min read
What's a "bless walk"

A bless walk is when someone hands you a stack of decisions and asks you to confirm each one rather than re-debate it. The decisions have been made; the bless is your check that the maker understood the situation correctly. If you spot a wrong-feeling one, you flag it and it goes back to debate. Otherwise you bless and move on. SL-5.3 already wrote these down — your job is just to confirm or push back.

Bless 1 · Verdict D9

Where do dispatch-worker and relay-writer belong?

The boundary cut — _meta-skills/ vs mission-control/
The setup
dispatch-worker and relay-writer were sitting in _meta-skills/ because they were workshop machinery. But they don't operate on other skills — they operate on chats. dispatch-worker spawns a chat from a handoff. relay-writer writes about what a chat is doing. The mismatch was bugging you.
On the table
C1 Operand-cut: _meta-skills/ = operates on skills. mission-control/ = operates on chats. ✓
C2 Audience-cut: workshop-universal vs domain-specific. Deferred to AL-0.
C3 Install-policy-cut: never-installs vs declared-per-suite. Used as enforcement, not the primary cut.
C4 Compositional-cut: atomic primitives vs workflows-with-conventions. Used as the tiebreaker on edge cases. ✓
Picked
C1 primary + C4 tiebreaker + C3 enforcement. Operand-cut is the simple rule. If something's ambiguous, ask "does it carry a convention?" — that's the tiebreaker. install-skill enforces by refusing.
On disk
Before _meta-skills/ scaffold-skill install-skill dispatch-worker relay-writer audit-skill red-team-skill build-skill After _meta-skills/ scaffold · install · audit · red-team · build · ... mission-control/ dispatch-worker relay-writer (+ log-relay, retro already here)
Two skills migrate. The other 13 meta-skills stay put.
What changes
The SL-7.1.3 migration handoff is now blessed and ready to run — it does the actual git mv work. red-team-skill stays in _meta-skills/ (it operates on a skill, even though it spawns a subagent). New workshop-discipline §18 codifies this cut.
Your bless

Does it still feel right that "operates on skills" vs "operates on chats" is the line? Or does that cut feel arbitrary when you say it out loud? If it feels off, the alternative on the table was audience-cut (workshop-universal vs domain-specific) — that one got deferred to AL-0 when Josh onboards. Bless or flag.

Bless 2 · Verdict D8

How does mission-control/ look inside?

Flat folders vs sub-grouping · per-skill _system/ or absorbed
The setup
mission-control is the first family workshop (Concept 05 in Lesson 1). When you've got 4–8 sibling skills sharing a domain, how do you arrange them? Flat? Sub-grouped by function? And does each skill keep its own _system/, or share one?
On the table
A1 Flat one-folder-per-skill. mission-control/log-relay/, mission-control/retro/, etc., no nesting. ✓
A2 Sub-grouped by function: launchers/, loggers/, dashboards/. Over-engineered now.
A3 Hybrid: flat until N≥5 same-cluster, then promote.
B3 Per-skill _system/ + family-wide _system/. Three levels: workshop, family, per-skill. ✓
Picked
A1 (flat) + B3 (three-level _system/). Don't over-organize. Each child skill keeps its own filing cabinet; the family folder gets its own for cross-skill stuff. Re-debate triggered if mission-control grows past 8 siblings.
Why not A2
Sub-grouping forces premature categorization. Is relay-writer a "logger" or a "convention-owner"? You'd have to pick. With flat + a Children table in CLAUDE.md, you can describe the role in words without committing the filesystem to it.
Your bless

The trigger says "re-debate at 8 siblings." mission-control currently has 2 (log-relay, retro) + 2 incoming (dispatch-worker, relay-writer) = 4. You're halfway to the trigger. Does N=8 feel like the right threshold? Too low? Too high? Bless or flag.

Bless 3 · Verdict D10 · DEFERRED

Where do experiment logs live?

Workshop-level vs family-level vs per-skill
The setup
SL-7.3.0 is a research chat specifically tasked with figuring out experiment conventions. SL-5.3 looked at the question and said "I'm not going to pick — the chat already owns this question." That's a deferral, not a punt. The decision still gets made, just by a different seat.
On the table
D1 Workshop-level skills-lab/_system/experiments/. Max discoverability, cross-cuts families.
D2 Family-level mission-control/_system/experiments/. Natural fit if all experiments are mission-control-domain.
D3 Per-skill <skill>/_system/experiments/. Max ownership, min visibility.
D4 Dedicated top-level experiments/ as peer to mission-control.
Picked
None — deferred to SL-7.3.0. Informational pre-stake from SL-5.3: probably D1 if experiments cross-cut multiple suites; probably D2 if they stay mission-control-domain. SL-7.3.0 picks; SL-5.3 ratifies (or pushes back) when the recommendation lands.
Your bless

You already launched SL-7.3.0. It's now authoritative on this question. When it returns a recommendation, SL-5.3 will hand you another bless card. For now: bless the deferral itself. Does it feel right that the question stays open until the research chat finishes? Or do you want to force a pick now? (Usually: bless the defer.)

Bless 4 · Verdict D11

Is "SL-7.X = mission-control" still the right naming?

Keep broad vs split off experiments vs split off taxonomy
The setup
SL-7.X has been the prefix for everything mission-control-flavored: SL-7.1 (dispatch-worker), SL-7.2 (relay-writer), SL-7.3 (experiments), SL-7.4 (this brainstorm). Three different shapes — build, research, brainstorm — all under one prefix. Worth splitting?
On the table
E1 Keep broad. Everything mission-control-domain is SL-7.X.
E2 Split: SL-EX.X for experiments. Cleaner shape; risks prefix sprawl.
E3 Split: SL-TX.X for taxonomy/brainstorm chats.
E4 Conservative hybrid. Keep broad for now. Spin off lineages only when empirically warranted. ✓
Picked
E4 — conservative hybrid. SL-7.X stays broad. Trigger to re-debate: 12+ SL-7 chats with 3+ of each shape (build / research / brainstorm). Until then, the interleaving is fine.
Your bless

Does "12 chats with 3+ of each shape" feel like the right re-debate trigger? Some operators prefer time-based triggers ("revisit at 6 months"). Some prefer pain-based ("only if it starts hurting"). SL-5.3 picked a volume/diversity threshold. Bless or flag.

Bless 5 · Verdict D12

Does mission-control stay yours, or get shared with Josh?

SL-scope (workshop-local) vs AL-scope (universal)
The setup
Josh is going to onboard. When he clones the monorepo, does mission-control/ stay workshop-local (each contributor builds their own), or does it get promoted to a universal shared layer? Or some hybrid?
On the table
F1 Workshop-scoped forever. Each contributor has their own mission-control. ✓
F2 Universal via AL-scope. One shared dispatch infrastructure. Deferred to AL-0.
F3 Hybrid: universal schema (relay format) + local implementation. Pre-staked as the most interesting long-term shape.
F4 REJECTED. Per-contributor workshop fork. Fights the monorepo decision.
Picked
F1 — workshop-scoped for now. Josh's mission-control can diverge from yours if you want. F2/F3 deferred to AL-0. Escalation trigger: when someone actually wants to dispatch INTO another contributor's chat, that's when F2/F3 gets ratified.
Why F4 is rejected
The monorepo decision (locked in monorepo-conventions.md §1) said one shared repo, both contributors push to main. F4 would mean each contributor maintains a hard fork — which violates that. SL-5.3 has authority to reject anything that fights an upstream lock.
Your bless

Two things to bless: (a) is workshop-scoped the right default for now? and (b) is "someone wants to dispatch INTO another contributor's chat" the right escalation trigger? The second one matters more — it decides when AL-0 has to weigh in. Bless or flag either.

SL-5.3 also flagged this

Across D8 (re-debate at N=8), D11 (re-debate at 12+ chats), and earlier verdicts, the same shape keeps appearing: "lock the current state as default; codify a re-debate trigger; promote on second instance." SL-5.3 suggested formalizing this pattern as a named convention — maybe convention-by-precedent. That's a future bless walk, not this one. Filed as a follow-up dispatch candidate.

Recap

Five blesses on the table.