docs(handoff): the leaving-off point — a fresh Fable agent continues from here #146

hkowalski · 2026-07-01T12:02:16+02:00

hkowalski commented

2026-07-01 12:02:16 +02:00

docs(handoff): the leaving-off point — a fresh Fable agent continues from here

Along the way:

docs(handoff): the orchestration workflow, in two lines
stopped test.sh from working, we need to save my p-cores
scripts: test-rotate can invoke test.sh again — the refusal now keys on CK_VIA_TEST_ROTATE
docs: record the owner's rulings — #10 bytes law stands, #11 tag DERIVED, B7 retired
docs(coherence): mark the #21 key contradiction RESOLVED — TypeArity[<=6] stands, cells are the answer

Touches 7 files: CLAUDE.md, docs/coherence.md, docs/decisions/decisions-backend-locks.md, handoff.md, questions.md, scripts/test-rotate.sh and 1 more.

docs(handoff): the leaving-off point — a fresh Fable agent continues from here Along the way: - docs(handoff): the orchestration workflow, in two lines - stopped test.sh from working, we need to save my p-cores - scripts: test-rotate can invoke test.sh again — the refusal now keys on CK_VIA_TEST_ROTATE - docs: record the owner's rulings — #10 bytes law stands, #11 tag DERIVED, B7 retired - docs(coherence): mark the #21 key contradiction RESOLVED — TypeArity[<=6] stands, cells are the answer Touches 7 files: `CLAUDE.md`, `docs/coherence.md`, `docs/decisions/decisions-backend-locks.md`, `handoff.md`, `questions.md`, `scripts/test-rotate.sh` and 1 more.

mdoorn approved these changes

2026-07-01 13:38:15 +02:00

mdoorn left a comment

Yes. The commit messages carried me through all 7 files, which is not always true.

handoff.md

					
				@ -84,2 +36,2 @@

				  repeated first-run-after-rebuild gate 101s at merges (all green on

				  re-run). A flaky gate is a quality defect and it is getting worse.

				## Next (dependency order, after in-flight merges)

				1. test-fn 3d: the Assert actor (roster check/true/equal), ratchet 4 → 0.

actor reads, decides and writes in one function. The deciding part is the piece I would want to test on its own. Worth splitting?

`actor` reads, decides and writes in one function. The deciding part is the piece I would want to test on its own. Worth splitting?

scripts/test.sh

					
				@ -16,0 +19,4 @@

				# cpu_broken.md) and the gate must run through test-rotate.sh, which offlines

				# nothing but pins the run to a rotating CPU subset and refuses if the bad

				# core is back online. test-rotate.sh sets the marker below when it is the

				# caller; nothing else may set it.

I had to read caller twice. Nothing wrong with it, but one line of comment would save the next person the second read.

I had to read `caller` twice. Nothing wrong with it, but one line of comment would save the next person the second read.

hkowalski commented

2026-07-01 14:33:07 +02:00

Split. The decision in actor is a pure function now, with its own tests in handoff.md.

Split. The decision in `actor` is a pure function now, with its own tests in `handoff.md`.

hkowalski commented

2026-07-01 14:46:50 +02:00

Comment added on caller in scripts/test.sh. It was obvious when I wrote it and it will not be in a month.

Comment added on `caller` in `scripts/test.sh`. It was obvious when I wrote it and it will not be in a month.

hkowalski added 6 commits

2026-07-01 16:36:32 +02:00

docs(handoff): the leaving-off point — a fresh Fable agent continues from here 0717410332

Everything landed, nothing in flight. Position per plan with hashes, the
dependency-ordered Next, every open owner decision, the machine's condition,
and the orchestration laws this session learned the hard way (base
verification, tree-truth counts, honest-boundary stops, the rotating gate).

docs(handoff): the orchestration workflow, in two lines 922bfde289

stopped test.sh from working, we need to save my p-cores 478c4b7c36

scripts: test-rotate can invoke test.sh again — the refusal now keys on CK_VIA_TEST_ROTATE 10bc6e3ba6

04145918 made test.sh exit 1 unconditionally to force the rotate wrapper,
but the wrapper calls test.sh, so the full gate had no working entry at
all. The refusal stays for direct callers; test-rotate.sh sets the marker.

docs: record the owner's rulings — #10 bytes law stands, #11 tag DERIVED, B7 retired 31c6dee27e

questions.md #1-23 now all ruled and acked inline. Sizes S1a-c UNGATED
(marker flips ride the slices that use each ruling). B7 marked RETIRED in
decisions-backend-locks.md, superseded by the DET-6 slot model. handoff.md
rewritten: two slices in flight in manually-seeded worktrees; the Agent-tool
auto-worktree stale-seed law recorded. CLAUDE.md motto line is the owner's.

docs(coherence): mark the #21 key contradiction RESOLVED — TypeArity[<=6] stands, cells are the answer fce243881d

mdoorn manually merged commit 8b054cc091 into master

2026-07-01 16:36:32 +02:00

mdoorn referenced this pull request from a commit

2026-07-01 16:36:32 +02:00

Merge pull request #146 from northgale/feat/146-docs-handoff-the-leaving-off-point-a-fresh-fab

Sign in to join this conversation.

No reviewers