Round 41 — OpenSpec backfill founding + operator-algebra cadence + in-round v2 supersedure + DORA substrate

[AceHack]

Summary

Round 41 — OpenSpec backfill founding + operator-algebra cadence ship + Viktor P0 close + memory-restructure design + ledger actualisation + Round-35 holdover close + skill-tune-up catch-up + router-coherence ADR + post-Arc-8 self-correction sweep + in-round v2 supersedure closing 10 Kira findings + grandfather-claims inventory honouring v2's within-round commitment + DORA 2025 reference substrate lands in-tree.

33 commits organised into 12 narrative arcs (see docs/ROUND-HISTORY.md Round 41 section):

• Arc 1 (d435126) — OpenSpec coverage audit surfaces ~93% of shipped F# surface lacks a capability spec; ADR docs/DECISIONS/2026-04-21-openspec-backfill-program.md commits to one-capability-per-round cadence with Viktor adversarial-audit gate.
• Arc 2 (e51ec1b) — First cadence ship: operator-algebra capability extended with F# profile (openspec/specs/operator-algebra/profiles/fsharp.md) naming all primitive operators, chain-rule helpers, and invariants.
• Arc 3 (92d7db2) — Viktor (spec-zealot) P0 close: four drift defects (namespace, phantom Reset, after-step scope, lifecycle phase undercount) closed in-capability.
• Arc 4 (56f34b5) — Viktor's 10 P1 findings filed as Round-42 absorb work under parent backfill entry in docs/BACKLOG.md.
• Arc 5 (6e6e211, 36797ba) — ROUND-HISTORY arcs 1-4 narrative + memory-folder role-restructure design plan at docs/research/memory-role-restructure-plan-2026-04-21.md (4 open sign-off questions; execution deferred to Round 42 under Auto Mode do-not-take-overly-destructive-actions clause).
• Arc 6 (85fb352) — BP-WINDOW ledgers for Rounds 37-40 actualised with (merged via PR #30, 1e30f8c) provenance — forecast → settled observation.
• Arc 7 (e461d9c, 15e9654) — Soraya (formal-verification-expert) tool-coverage audit on RecursiveSigned.fs skeleton closes the Round-35 holdover gate; four tool-coverage prereqs captured in BACKLOG at priority-ordered granularity (TLC CI wire-up, Z3 S2 lemma, FsCheck S4 property, FsCheck refinement cross-check).
• Aarav catch-up (085c0e3) — skill-tune-up ranking re-run after 23-round cadence gap (ranker offline rounds 19-40); round-18 HAND-OFF-CONTRACT carry-over re-entered top-5 rather than silently dropped.
• Arc 8 (47d92d8) — Router-coherence ADR v1 docs/DECISIONS/2026-04-21-router-coherence-claims-vs-complexity.md codifies analytic-first → empirical-second pipeline between Hiroshi (complexity-reviewer) and Daisy (claims-tester), closing Aarav's 23-round-stale round-18 finding.
• Arc 9 (d76a09b, 2042a85, fcfa3d9, 779d7ef) — Post-Arc-8 self-correction sweep: Prereq 1 re-sized S→M after discovering no TLC CI job exists (22-round compile-only drift on RecursiveCountingLFP.tla); formal-analysis-gap-finder round-42 entry filed as recurring audit lens for "verifier-installed ≠ verifier-runs" across every spec directory; CONFLICT-RESOLUTION.md Hiroshi ↔ Daisy row closes Arc-8's canonical-authority-matrix follow-up; Kira (harsh-critic) adversarial pass over ADR 47d92d8 surfaces 3 P0 + 5 P1 + 2 P2 findings, filed as round-42 supersedure item. Pattern: Arc 8 corrects Aarav's round-18 finding; Arc 9 catches Arc 8 itself under-reviewed.
• Arc 10 (09f0889, 4efe545, 4537365) — In-round v2 supersedure: docs/DECISIONS/2026-04-21-router-coherence-v2.md supersedes v1 in the same round, closing all 10 Kira findings via named textual closures C-P0-1 through C-P2-10 (grandfather clause bounded, reverse trigger unconditional, escalation-evidence loop unblocked, Status aligned, trigger list widened, decision-table author-bounce explicit, escalation timebox added, Kenji named as binding dispatcher, BCL-contract example fixed, scope-creep cut). v1 retained with Superseded-by header per GOVERNANCE §2 so CONFLICT-RESOLUTION.md citation chain stays resolvable. BACKLOG supersedure entry discharged in-round. Pattern: Arc 9 surfaces the under-review; Arc 10 lands the close in the same round rather than deferring the known-imperfect artefact to round-42.
• Arc 12 (46075d6, 0111e3f) — DORA 2025 reference substrate lands in-tree: two untracked PDFs (docs/2025_state_of_ai_assisted_software_development.pdf 15MB / 138pp, docs/2025_dora_ai_capabilities_model.pdf 9MB / 94pp; CC BY-NC-SA 4.0) that had been sitting in the working tree since 2026-04-20 02:13/02:17 now committed as the in-repo substrate behind the 2026-04-20 memory promotion of DORA from external-anchor to measurement-frame starting point. Arc narrative surfaces the /next-steps ranker-scope gap (nine re-fires coasted on in-context summary rather than checking git status --short) as candidate skill-tune-up work for Aarav. In-tree citation anchors now reachable: Nyquist stability criterion (p9 fn 1) for CI-meta-loop + retractable-CD P1, "AI is an amplifier" for corporate-religion / sandbox-escape threat class, seven-capability model for external-audience round-audit vocabulary.
• Arc 11 (d98ef2b) — Grandfather-claims inventory honours v2 Closure C-P0-1's within-round commitment: docs/research/grandfather-claims-inventory-2026-04-21.md catalogues 35 live O(·) claims at ADR-landing time (29 F# /// docstrings, 3 grey-zone F# code comments, 1 openspec/specs/operator-algebra/spec.md line, 2 docs/research/** claims). Distinguishes live claims from historical evidence. Matching docs/BACKLOG.md P2 discharge entry: one-claim-per-round cadence, ~35-round tail, Aarav graceful-degradation clause fires if ≥3 consecutive rounds pass without discharge. Pattern: without Arc 11, Arc 10 would have shipped a contract Zeta didn't meet.
• BP-WINDOW ledger rows (098fa25, 7c4b01a, 459b218, 47b82d7, 531dc68, ec00be9, 1e31d34, 913d054, 7b581da) — per-commit window-expansion accounting; meta-ledger-only commits do not get self-referential rows (established stopping rule).

Test plan

• dotnet build -c Release — 0 Warning(s), 0 Error(s) (verified this session)
• BP-10 invisible-Unicode audit on all round-41 files including v2 ADR — clean
• Per-commit BP-WINDOW ledger rows present for every primary-work commit (arc count = 12, primary-commit count = 12, alignment restored)
• ADR follow-up loop closed at docs/CONFLICT-RESOLUTION.md (Hiroshi ↔ Daisy Active-tensions row pointing at v1 — supersedure-header redirect to v2 keeps the citation chain live)
• Harsh-critic adversarial review of round-41 ADRs dispatched; v1 findings closed in-round via v2 supersedure (all 10 findings carry named textual closures in v2); audit trail preserved in BACKLOG closure annotation

Round-42 carried work

• Memory-folder role-restructure execution (Phase 1 of 5) — gated on 4-question sign-off
• Soraya's Prereqs 1-4 (TLC CI + Z3 + FsCheck) — M-effort; Prereq 1 now correctly sized
• Viktor's 10 P1 findings absorb into operator-algebra spec
• ADR v2 follow-up SKILL.md edits via skill-creator (claims-tester + complexity-reviewer) — targets v2 as intended (no v1-vs-v2 sequencing dependency anymore since v2 landed in-round)
• Grandfather O(·) claims discharge — one claim per round through the v2 pipeline (35-round tail; P2 BACKLOG entry; Aarav graceful-degradation clause)
• lsm-spine-family capability draft (ADR round-42 slot)
• formal-analysis-gap-finder round-42 invocation (scheduled after Prereq 1 lands so audit sees corrected state)
• Aarav's top-5 tune-up actions via skill-creator (now unblocked — both the two v1-gated items now target v2)

[Copilot]

Pull request overview

Establishes the Round 41 OpenSpec backfill program and captures new/updated governance + research artefacts to support delete-recovery specs and router-coherence (analytic ↔ empirical) workflow, plus lands DORA 2025 reference PDFs in-tree.

Changes:

• Extends the operator-algebra OpenSpec with lifecycle, scheduling, clock-scope, chain-rule wrapper, and Z-set representation requirements.
• Adds/updates ADRs, BACKLOG, CONFLICT-RESOLUTION, ROUND-HISTORY, and research notes to formalize the OpenSpec cadence and router-coherence v2 supersedure.
• Updates persona notebooks/scratchpads and adds DORA 2025 PDFs as reference substrate.

Reviewed changes

Copilot reviewed 14 out of 16 changed files in this pull request and generated 8 comments.

Show a summary per file

File	Description
openspec/specs/operator-algebra/spec.md	Adds new operator-algebra requirements and scenarios (lifecycle, strictness, clock scopes, incremental wrapper, Z-set invariants).
openspec/specs/operator-algebra/profiles/fsharp.md	Aligns the F# profile namespace/assembly to Zeta.Core / Zeta.Core.
memory/persona/soraya/NOTEBOOK.md	Records Round 41 tool-coverage audit for RecursiveSigned and prereq routing.
memory/persona/best-practices-scratch.md	Refreshes scratchpad formatting and appends Round 41 live-search findings.
memory/persona/aarav/NOTEBOOK.md	Adds Round 41 catch-up ranking and updated top-5 tune-up targets.
docs/research/openspec-coverage-audit-2026-04-21.md	Introduces an OpenSpec coverage audit and proposed cadence plan.
docs/research/memory-role-restructure-plan-2026-04-21.md	Design plan for restructuring memory/ by role, including execution steps.
docs/research/grandfather-claims-inventory-2026-04-21.md	Inventories pre-ADR O(·) claims to seed the v2 router-coherence discharge cadence.
docs/ROUND-HISTORY.md	Adds the Round 41 narrative and updates ledger headers for rounds 37–40 provenance.
docs/DECISIONS/2026-04-21-router-coherence-v2.md	Adds v2 ADR superseding v1 and closes harsh-critic findings via explicit closure clauses.
docs/DECISIONS/2026-04-21-router-coherence-claims-vs-complexity.md	Adds v1 ADR (now marked superseded) documenting the initial hand-off pipeline.
docs/DECISIONS/2026-04-21-openspec-backfill-program.md	Adds ADR establishing one-capability-per-round OpenSpec backfill cadence and audit gate.
docs/CONFLICT-RESOLUTION.md	Adds an Active-tensions row for Complexity Reviewer ↔ Claims Tester with a standing resolution pointer.
docs/BACKLOG.md	Rewrites OpenSpec backfill entry with Round 41 status, adds/router-coherence + grandfather discharge tracking, and expands memory-role restructure entry.
docs/2025_state_of_ai_assisted_software_development.pdf	Adds DORA 2025 reference PDF to the repo.
docs/2025_dora_ai_capabilities_model.pdf	Adds DORA capabilities model PDF to the repo.

[Copilot]

The Standing Resolution link points to the v1 ADR (router-coherence-claims-vs-complexity.md), which is now explicitly superseded by v2. Since this row is meant to be the current operating contract, it should link directly to docs/DECISIONS/2026-04-21-router-coherence-v2.md (and optionally mention that v1 is retained as historical record).

Suggested change

	`docs/DECISIONS/2026-04-21-router-coherence-claims-vs-complexity.md`
	— analytic-first on new claims; reverse trigger (Claims Tester
	`docs/DECISIONS/2026-04-21-router-coherence-v2.md` —
	analytic-first on new claims; reverse trigger (Claims Tester

[Copilot]

The Round 41 narrative says operator-algebra/spec.md is now 324 lines, but the file is currently 365 lines. Please update this line-count so the round history stays internally consistent with the repo state.

Suggested change

	324 lines. F# profile at `openspec/specs/operator-algebra/
	365 lines. F# profile at `openspec/specs/operator-algebra/

[Copilot]

This doc links to docs/research/openspec-coverage-audit-2026-04-21-inventory.md, but that file does not appear to exist in the repo/PR. Either add the referenced inventory file or adjust the text/link to point at the actual location of the full 66-row table.

Suggested change

	See `docs/research/openspec-coverage-audit-2026-04-21-inventory.md`
	(follow-up) for the full 66-row table. Summary by band below;
	the table and Band definitions are this document's load-bearing
	artefact.
	Summary by band below; the Band definitions in this document
	are the load-bearing artefact.

[Copilot]

The heading for the uv-only Python candidate is accidentally split into two separate H2 headings ("## … package and tool" then "## management"), which will break the section structure/ToC. Combine this into a single heading line (or make the second line plain wrapped text).

[Copilot]

The Phase 1 grep commands walk the whole repo (".") but don't exclude references/upstreams/**, which the repo explicitly warns against for recursive scans because it contains many full upstream clones. Add an exclude (e.g., --exclude-dir=references / --glob '!references/upstreams/**') to these and the other grep loops in this plan so the counts/runtime are meaningful.

Suggested change

	grep -rln "memory/persona/" --include="*.md" --include="*.json" \
	--include="*.sh" --include="*.fs" --include="*.cs" . | \
	grep -v "^./\\.git" | wc -l # → 114 files
	grep -rc "memory/persona/" --include="*.md" --include="*.json" \
	--include="*.sh" . | grep -v ":0$" | \
	awk -F: '{s+=$2} END {print s}' # → 700 refs
	grep -rln "memory/persona/" --exclude-dir=.git --exclude-dir=references \
	--include="*.md" --include="*.json" --include="*.sh" \
	--include="*.fs" --include="*.cs" . | wc -l # → 114 files
	grep -rc "memory/persona/" --exclude-dir=references \
	--include="*.md" --include="*.json" --include="*.sh" . | \
	grep -v ":0$" | awk -F: '{s+=$2} END {print s}' # → 700 refs

[Copilot]

The crosswalk lists memory/persona/README.md moving to memory/persona-README.md (or memory/README-persona-roles.md), but Phase 2 later uses memory/persona-roles-README.md. Please pick one canonical target name and make the crosswalk + git mv commands consistent to avoid broken references during execution.

Suggested change

	memory/persona/README.md memory/persona-README.md (or moved to memory/README-persona-roles.md)
	memory/persona/README.md memory/persona-roles-README.md

[Copilot]

The Phase 3 rewrite loop mixes BSD- and GNU-specific flags (sed -i "" is BSD/macOS, but xargs -r is GNU-only and fails on macOS). Either adjust the commands to be portable (or provide both macOS/Linux variants) so the plan can be executed reliably.

Suggested change

	xargs -r sed -i "" "s|memory/persona/$persona/|memory/$role/$persona/|g"
	done
	# Clean up the flat-file references too
	grep -rl "memory/persona/README\\.md" --include="*.md" . | \
	xargs -r sed -i "" "s|memory/persona/README\\.md|memory/persona-roles-README.md|g"
	grep -rl "memory/persona/best-practices-scratch\\.md" \
	--include="*.md" --include="*.sh" . | \
	xargs -r sed -i "" \
	"s|memory/persona/best-practices-scratch\\.md|memory/best-practices-scratch.md|g"
	while IFS= read -r file; do
	sed -i.bak "s|memory/persona/$persona/|memory/$role/$persona/|g" "$file" && rm -f "$file.bak"
	done
	done
	# Clean up the flat-file references too
	grep -rl "memory/persona/README\\.md" --include="*.md" . | \
	while IFS= read -r file; do
	sed -i.bak "s|memory/persona/README\\.md|memory/persona-roles-README.md|g" "$file" && rm -f "$file.bak"
	done
	grep -rl "memory/persona/best-practices-scratch\\.md" \
	--include="*.md" --include="*.sh" . | \
	while IFS= read -r file; do
	sed -i.bak \
	"s|memory/persona/best-practices-scratch\\.md|memory/best-practices-scratch.md|g" \
	"$file" && rm -f "$file.bak"
	done

[Copilot]

This research doc uses the human maintainer’s name ("Triggering question: Aaron …"). Repo convention is to keep the human maintainer’s name out of general docs and use the role reference ("human maintainer") instead. Please rewrite this (and similar occurrences in docs/research/**) to use the role wording to match .github/copilot-instructions.md ("No name attribution…").

Suggested change

	**Triggering question:** Aaron 2026-04-20 — *"opensepcs, if I
	**Triggering question:** human maintainer, 2026-04-20 — *"opensepcs, if I

[AceHack]

Classification (per docs/AGENT-GITHUB-SURFACES.md § Surface 1)

Shape stack (higher-in-list wins on multi-match):

• behind-main — mergeStateStatus: BEHIND
• has-review-findings — Copilot left 8 actionable findings
• awaiting-human — Aaron-scoped decision

Relationship to #32. round-42-speculative descends from round-41's tip (git merge-base confirms); merging #32 would subsume this PR. This is the Aaron-scoped part: does #31 merge as a smaller slice, or close in favour of #32?

Copilot findings triage (8 actionable):

• 3 simple fixes: ADR v1→v2 link, line-count mismatch, jq quoting — can be addressed under this PR if it merges.
• 4 author-edits in research docs — low risk, author judgement.
• 1 cross-repo xref to a file that doesn't exist — needs author decision (drop xref or land target).

Mirrored to docs/hygiene-history/pr-triage-history.md (bootstrap + row for this PR) and docs/HUMAN-BACKLOG.md pending Aaron's call.

Classification taxonomy is declared non-final per the beef-up clause — first 5-10 rounds of fire-history feed a revision pass.

🤖 Per the new umbrella playbook. Context is in docs/AGENT-GITHUB-SURFACES.md + .claude/skills/github-surface-triage/SKILL.md.

[Copilot]

Pull request overview

Copilot reviewed 5 out of 5 changed files in this pull request and generated 4 comments.

[Copilot]

The narrative claims the operator-algebra/spec.md is “now at 365 lines”, but the current file openspec/specs/operator-algebra/spec.md is 679 lines. Please update the number (or clarify what subset is being counted) so this cross-reference stays accurate.

[Copilot]

Phase 3’s grep -rl search/replace loop doesn’t exclude the top-level references/ directory. In this repo that directory is intentionally huge, so this command will be very slow (and potentially noisy) even though later pipeline stages filter paths. Add --exclude-dir=references (and keep it consistent with Phase 1’s pre-flight greps) to avoid traversing ./references/** during the rename.

[Copilot]

This grep -rl (README pointer cleanup) should also exclude references/ for the same reason as the Phase 1 greps—otherwise it will traverse the large ./references/** tree even though the replacements only target repo-local docs.

[Copilot]

Same as above: this grep -rl for best-practices-scratch.md references should exclude references/ (e.g., --exclude-dir=references) to avoid an expensive recursive walk over vendored upstream clones during the rename.

Suggested change

	--include="*.md" --include="*.sh" . | \
	--include="*.md" --include="*.sh" --exclude-dir=references . | \