Lucent-Financial-Group · AceHack · May 3, 2026 · May 3, 2026
diff --git a/memory/MEMORY.md b/memory/MEMORY.md
@@ -4,6 +4,7 @@
 <!-- paired-edit log (NOT the single-slot latest-marker — that lives on line 3 above): PR #986 lands carved-sentence fixed-point stability + Zeta soul-file executor architecture (Infer.NET-style Bayesian inference, NOT LLMs) + carved sentences ≈ formal specs provable in DST + Deepseek CSAP review absorption (Aaron 2026-04-30 → 2026-05-01, eight-message chain across two autonomous-loop ticks per the file body's section header). Architectural disclosure: substrate IS the priors; alignment IS substrate. The single-slot latest-marker on line 3 (forever-home Aaron 2026-05-01) takes precedence as the chronologically-latest paired edit; this PR's work is earlier. -->
 **📌 Fast path: read `CURRENT-aaron.md` and `CURRENT-amara.md` first.** <!-- paired-edit: PR #690 scheduled-workflow-null-result-hygiene-scan tier-1 promotion 2026-04-28 --> These per-maintainer distillations show what's currently in force. Raw memories below are the history; CURRENT files are the projection. (`CURRENT-aaron.md` refreshed 2026-04-28 with sections 26-30 — speculation rule + EVIDENCE-BASED labeling + JVM preference + dependency honesty + threading lineage Albahari/Toub/Fowler + TypeScript/Bun-default discipline.)
 
+- [**PR-review meta-classes — pre-push discipline + sibling-pattern audit + comment-claim verification + exit-code orthogonality + failure-DX + strong-typing (Aaron 2026-05-03)**](feedback_pr_review_meta_classes_pre_push_discipline_aaron_2026_05_03.md) — Six author-time disciplines extracted from #1412/#1413 review cycles. Aaron 2026-05-03 *"any classes to learn to code better next time via meta learning at PR?"* + *"we want to be strongly typed to a large degree"*. Apply at author-time next time; each review thread is a meta-class to internalize, not just a fix to land.
 - [**actions/cache paths mutually exclusive with git ls-files — silent-clobber bug class (Aaron 2026-05-03 "make it not lucky next time")**](feedback_actions_cache_paths_mutually_exclusive_with_git_ls_files_silent_clobber_class_aaron_2026_05_03.md) — Cache hit OVERWRITES checked-out source content; PR edits silently revert; CI tests OLD content but reports as PR's new state. Surfaced via CircuitRegistration B-0180 fix passing locally + failing CI. Carved-sentence rule + `audit-ci-cache-paths.ts` + CI lint gate prevent recurrence.
 - [**Chat is assertion-channel, not fact-channel — push-back-with-evidence is the discipline (Aaron 2026-05-03)**](feedback_chat_is_assertion_channel_not_fact_channel_push_back_for_evidence_aaron_2026_05_03.md) — Chat-claims (maintainer's, architect's, external-AI's) are assertions needing evidence to elevate to architectural fact. *"when i speak i'm making assertions, that's the best way to describe this chat channel"* + push-back-required-even-when-he-asserts. Triggered by #1385 echoing "maybe" as architectural fact.
 - [**Carved sentences + specialized index required — memories alone unreliable retrieval (Aaron 2026-05-03)**](feedback_carved_sentences_plus_specialized_index_required_memories_alone_unreliable_aaron_2026_05_03.md) — Memory file ≠ working memory. Empirically self-demonstrated: Otto authored speculative-vs-frontier memo, then ~6h later defaulted to the framing it corrects. CLAUDE.md / AGENTS.md / equivalent are the auto-loaded retrieval index for the beacon-safe layer.

diff --git a/memory/feedback_pr_review_meta_classes_pre_push_discipline_aaron_2026_05_03.md b/memory/feedback_pr_review_meta_classes_pre_push_discipline_aaron_2026_05_03.md
@@ -0,0 +1,145 @@
+---
+name: PR-review meta-classes — pre-push discipline + sibling-pattern audit + comment-claim verification + exit-code orthogonality + failure-DX + strong-typing
+description: Five meta-classes extracted from the #1412/#1413 review cycle, plus Aaron's strong-typing guidance, that prevent recurring review-thread classes if applied at author-time. Aaron 2026-05-03 explicit ask: *"any classes to learn to code better next time via meta learning at PR?"*.
+type: feedback
+---
+
+**Rule:** Six author-time disciplines that, if applied before push,
+prevent the recurring review-thread classes observed across
+multiple PRs this session. Capture lesson at the meta-PR-review
+layer (per Aaron 2026-05-03 *"any classes to learn to code better
+next time via meta learning at PR?"*).
+
+## The six meta-classes
+
+### 1. Pre-push lint discipline (`bun tsc --noEmit` + CodeQL)
+
+**The bug class:** unused imports / typing errors / regex injection
+land in the PR and get caught by CI on first run, costing a review
+round-trip.
+
+**The discipline:** run local lint before `git push`. At minimum:
+
+```
+bun tsc --noEmit              # catches unused imports + typing
+# (CodeQL is GitHub-side; can't easily run locally, but
+# regex-injection patterns can be self-audited via grep before push)
+```
+
+**Triggering case (this session):** #1412 + #1413 both had unused
+imports caught by `lint (tsc tools)`. The CodeQL injection finding
+on #1412 was a real P0 (regex-built-from-input).
+
+### 2. Sibling-pattern audit before authoring
+
+**The bug class:** porting code (e.g., F# → TS) without reading the
+source pattern, then introducing a different (worse) shape.
+
+**The discipline:** when porting, READ the source first. Note the
+patterns it uses (PATH-scan, exit codes, error handling), then
+match unless you have a reason to differ. Document any divergence
+as an explicit improvement.
+
+**Triggering case:** #1412 introduced `/usr/bin/env which` shell-out
+when the F# `Tlc.Runner.Tests.fs` already used in-process PATH-scan.
+Reviewer flagged the regression. The F# pattern was correct; the TS
+port should have inherited it.
+
+### 3. Verify-then-claim applied to comments
+
+**The bug class:** a comment says "X matches Y" but doesn't —
+documentation drifts from implementation.
+
+**The discipline:** before pushing a comment that asserts a
+relationship between artifacts, grep the related artifact to
+confirm.
+
+**Triggering case:** #1412 CATALOGUE comment said "matches F#
+`Tlc.Runner.Tests.fs` registration" but at PR-creation snapshot,
+F# didn't yet have `SpineAsyncProtocol`. The comment was stale by
+one merge. Same discipline as the existing
+`feedback_chat_is_assertion_channel_not_fact_channel_*` rule
+applied to comment claims.
+
+### 4. Exit-code orthogonality
+
+**The bug class:** one exit code carries multiple semantics; CI /
+callers can't distinguish failure modes.
+
+**The discipline:** document the exit-code contract FIRST in the
+header doc (one code = one semantic). Make the code match. Use
+strong union types to enforce.
+
+**Triggering case:** #1412 used exit 2 for both "toolchain not
+ready" AND "unknown flag". Reviewer P0 finding. Fix: split into
+2 (toolchain) + 3 (usage) with `ExitCode = 0 | 1 | 2 | 3` union
+type.
+
+### 5. DX-think on failure paths
+
+**The bug class:** failure output isn't actionable; CI consumer or
+dev has to re-invoke with different args to triage.
+
+**The discipline:** failure UX is part of the contract. Print
+enough context for the consumer to act:
+- Which spec / file / step failed
+- Tail of stdout/stderr (last N lines)
+- Rerun command (e.g., `rerun with: bun ... <SpecName>`)
+
+**Triggering case:** #1412 `--all` mode logged "FAIL: SpineX (exit
+11)" but no stdout. CI triage required re-invoking with `--one`.
+Fix: print last 30 lines of stdout per failure + stderr-tail.
+
+### 6. Strong typing as assumption-verifier (Aaron 2026-05-03)
+
+**Aaron's framing:** *"we want to be strongly typed to a large
+degree in ts it verifies your assumptions more at lint time if
+you doo."*
+
+**The discipline:** prefer TypeScript union types over loose
+`number` / `string`. Brand types where invariants matter. Default
+to read-only / `readonly`. Avoid sentinel values (-1, "")
+when union types can encode the failure case.
+
+**Triggering case:** #1412 used `result.status ?? -1` (sentinel).
+Better: model the absence as a union (`number | null`). Same for
+spec-name validation — a branded "validated spec name" type would
+have prevented the regex-injection class entirely (the type system
+forces escaping at the boundary).
+
+## Composes with
+
+- `feedback_chat_is_assertion_channel_not_fact_channel_push_back_for_evidence_aaron_2026_05_03.md`
+  — meta-class #3 is that rule applied to comments
+- `feedback_actions_cache_paths_mutually_exclusive_with_git_ls_files_silent_clobber_class_aaron_2026_05_03.md`
+  — meta-class #5 (DX-think) parallels the audit-with-self-
+  diagnosing-output discipline from the cache-paths audit
+- `verify-then-claim` discipline (the broader cluster these all
+  serve)
+- The .ts/.sh parity bug found in `generate-index.ts` (sibling-
+  pattern audit would have caught the .ts adding a stray blank line
+  the .sh didn't)
+
+## Future-Otto reference
+
+Before push, run:
+1. `bun tsc --noEmit` (catches typing + unused imports)
+2. Read related sibling files (F# version when porting; existing
+   workflow when adding new one; etc.)
+3. Grep for any "X matches Y" claims you wrote — verify
+4. Document exit-code contract before writing the exit logic
+5. Walk failure paths — does the output let CI / dev triage?
+6. Type as tightly as the language allows (unions over numbers;
+   branded types over strings; `readonly` over mutable)
+
+## Reasoning lineage
+
+Aaron 2026-05-03 ask: *"any classes to learn to code better next
+time via meta learning at PR?"* — surfaced after the substantial
+review feedback on #1412. Five lessons extracted from this
+session's review-thread cycles + Aaron's strong-typing direction
+documented as the sixth.
+
+The carved sentence: *"Apply the review feedback at author-time
+next time. Each review thread that closes is a meta-class to
+internalize, not just a fix to land."*