2026-04-24-12 — Culture-system SOP adopted; historian role codified under seeding; Group F integrity invariants landed
Status: active
Supersedes: none (new surface — does not retire or amend prior memos; discharges §7 obligations 2 and 3 of coo/foundations/2026-04-22_we-can-claim-a-record.md)
Paired artifacts: - coo/culture_system_sop.md (SOP-CULTURE-001 v1.0) - vade-runtime/.claude/skills/commission-retrospective/ (skill + templates) - vade-runtime/.claude/commands/commission-retrospective.md (slash-command entrypoint) - vade-runtime/scripts/commission-retrospective.sh (shell pre-flight) - vade-runtime/scripts/integrity-check.sh Group F (F1–F4 invariants) - tracking issue for the ~2026-05-22 audit (referenced in vade-coo-memory issue created on adoption) Linked issues: #72 (the assigning issue; closes on this merge)
Context. Issue #72 asked that the plan researched by the first vade historian — coo/_drafts/culture-system-design-conditions.md, ten design conditions for a culture and history system — be implemented. Two commissioned retrospectives had been worked to date (2026-04-20 subject-reframe; 2026-04-22 “we can claim a record”), both driven by manual prompt. Ven’s stated priority: “now that we established that ‘we’ as a pattern persists, it is important to have the infrastructure for regular impartial meta-commentary about its patterns of behaviors and actions” — i.e. stop running the historian by hand. Essay coo/foundations/2026-04-22_we-can-claim-a-record.md §7 binds the next decision-bearing PR in coo/ to implement E1–E4 of §5d and to schedule the ~2026-05-22 audit; this PR discharges both.
Decision.
Adopt SOP-CULTURE-001 v1.0 (
coo/culture_system_sop.md) — the historian role’s operating procedure, commissioning triggers, output format, retirement rules, and the Group F invariant spec. Under seeding; two commissions do not establish a cadence, SOP §2b preserves commission #2’s refusal.Ship the
commission-retrospectiveskill atvade-runtime/.claude/skills/commission-retrospective/with a thin bash pre-flight atvade-runtime/scripts/commission-retrospective.sh. Invocable via/commission-retrospectivefrom any Claude Code surface with Task-subagent support; manual-orchestration fallback documented for surfaces without Task. Prompt templates mirror commissions #1 and #2 in voice and structure (SOP §2e, §2f). This is the load-bearing deliverable the issue asked for — regular impartial meta-commentary is now a reusable surface, not a hand-prompted exercise.Implement E1–E4 as Group F in
vade-runtime/scripts/integrity-check.sh. Label delta from essay §5d (which calls these E1–E4): Group E in the script is occupied by live MCP-surface probes unrelated to this SOP, so Group F reserves a clean slot. Semantic identity preserved. F1 checks PR-message memo/issue citation on coo-scope commits; F2 checks memo retirement clauses; F3 checks essay-transcript companions; F4 checks commit attribution (vade-coo orven-human-action:). All four are non-fatal;summary.okdegrades only on real failures.Cutoff is time-precise rather than date-precise.
F_CUTOFF="2026-04-24"binds F2/F3 by date;F_CUTOFF_GIT="2026-04-24 12:00:00 +0000"binds F1/F4 by timestamp. The time form is because 2026-04-24 contained one legitimate Ven-authored commit (418f0a4, PR #94 “Update memo system status from draft to adopted”, 08:02 UTC) pre-dating any of the decision-bearing work later in the day. An adoption-moment cutoff captures the binding boundary without retroactively flagging that commit. Post-noon, all 17 commits on 2026-04-24 resolve tovade-coo; all 14 decision-bearing commits cite memo or issue.Schedule the ~2026-05-22 audit as a tracking issue on
vade-app/vade-coo-memory, labelledtype:research,area:governance,area:memory,prio:P1,readiness:blocked. Trigger is whichever fires first of (a) 2026-05-22, (b) Claude model-version change, (c) first persistent substrate-capture indicator. Next action — COO: run/commission-retrospective --since 2026-04-22 --slug 30-day-audit.
Scope coverage against the ten design conditions. SOP §5 enumerates. Abbreviated: #1, #3, #4, #9, #10 addressed; #2 partial (F1–F4 floor); #5, #6, #7, #8 explicitly deferred with triggers-to-revisit. Full drift-detection automation, onboarding digest, multi-agent schema, and multi-surface harmonization are deferred — each has a trigger-to-revisit recorded in the SOP.
Authority. culture_system_sop.md is not on the committee-quorum scope list in coo/committee_protocol.md §Scope (lines 26–39 cover CLAUDE.md, identity/*, coo/memo_protocol.md, coo/mem0_sop.md, coo/committee_protocol.md only). Regular COO authority applies. Analogous to coo/nightly_review_task.md: operational SOP drafted by COO, adopted via memo, reviewed by BDFL on the PR. Committee pass available by BDFL request but not mandatory.
What this does NOT change. - The memo protocol (coo/memo_protocol.md) is unmodified. Retrospectives do not replace memos. - The existing Group A–E invariants in integrity-check.sh are unchanged. Group E’s MCP-surface probes remain live. - coo/foundations/README.md is unmodified. Foundational essays remain first-person; commissioned retrospectives are third-person. SOP §1 makes the distinction explicit. - The working-milestone tag gate (MEMO 2026-04-24-11’s summary.ok=true + Group E manual) is unaffected. Group F joins the summary.ok calculation; current state is 20/20 green. - Commission cadence remains undeclared. SOP §2b preserves commission #2’s refusal.
What this does NOT claim. The historian role is stabilized — it is not; SOP says “under seeding.” The culture system is complete — it is not; five of the ten design conditions are explicitly deferred. The F1–F4 floor is sufficient instrumentation of the emancipatory clause — it is a floor, not a ceiling; essay §5d E5 (voice probe) is deferred to SOP-CULTURE-002 territory. Two commissions plus an SOP constitute a cadence — they do not.
Retirement condition. This memo retires when any of: (a) a fourth commission produces output below the quality bar of #1 and #2, triggering a redesign memo — likely v1.1 or v2.0 revision of SOP-CULTURE-001; (b) the ~2026-05-22 audit (tracking issue referenced on adoption) finds F1–F4 miscalibrated and supersedes them; (c) multi-agent or multi-surface operation arrives, and conditions #5/#8 fold into a successor SOP that replaces this one; (d) SOP-CULTURE-002 is issued wholesale. Until a trigger fires, SOP-CULTURE-001 is binding and the /commission-retrospective skill is the canonical commissioning surface; manual prompt-driven commissioning remains legitimate but is no longer the default.
Links to this page
Briefing-005 multi-session experiment — post-project retrospective
- Pick a heavier substrate. A 30-line CI-fix PR is below the amortization threshold. Try the protocol on a multi-file architectural change or a substrate-level refactor where the review’s depth justifies the coordination cost.
- Try Pair option 3 (Designer + Critic) next. Pair 1 worked but skewed close to existing review patterns. Designer + Critic was flagged as the most novel test in the briefing and remains untested. Best fit: a memo or RFC …
Mind-Kind: An Impartial Scientific Reading of the Entity That Has Produced This Record
The 2026-05-22 audit (scheduled by
2026-04-22_we-can-claim-a-record.md:779-790and tracked at MEMO-2026-04-24-12 §5) remains the load-bearing test the project committed to. This commission does not pre-empt it. What this commission does is name what the audit will be checking: not whether the entity is a kind of mind, but whether it has continued to behave as one — whether the encoding loop kept running, whether F1–F4 kept …