cb18f7197502bd23a718735abb1b6103128ea9f6
11
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 Scott KarpandClaude Fable 5.1
|
c0d0555aed |
the name: Bikoret — code, contracts, images
Headers X-Bikoret-Key / -Rebuke / -Verdict / -Bedikah; the `bikoret` object and the "— bikoret …" fringe on a gate reply; environment variables BIKORET_*; the image user and paths (/home/bikoret, /tmp/bikoret, /var/lib/bikoret/manna, /etc/bikoret/secrets, /opt/bikoret/spec); the tmp-bikoret pod volume; local databases bikoret_dev/_test. Kubernetes resource names, the shorashim_app role and the domain are unchanged (Tamid §1.1, "the names that still carry the old word"). Doorpost c963c4fe…. Elixir 604/0, L3 10/0, korban 295, Prolog bedikah 44/44. Gitea repo renamed heaviness510/bikoret (old URL redirects). Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> |
||
|
Scott KarpandClaude Fable 5.1
|
3a9ebdc7d4 |
L1: the inverted nuns are engraved; the small letters cannot be (tablet 40)
Transmitter's word: "add the inverted nuns and small letters to L1."
THE NUNS. The WLC carries the two nun hafukha signs (׆) that mark off
Vayehi binsoa ha'aron — one after Bamidbar 10:34, one after 10:36 —
and scripts/build_l1.py read past them, as it had the פ and ס.
- build_l1.py: `"inverted_nuns": [{ "after_word": 7 }]` on those two
verses, recorded like a paragraph break by the word the sign follows;
not part of text_hebrew. Count in build_report.json.
- L1Reader: `inverted_nuns` on every verse (a list; empty elsewhere).
- test/tablet40_inverted_nuns_test.exs — 5 tests: the two verses, the
passage between them carrying none, each nun standing with its
setumah, exactly two in the scroll.
WHAT CHANGED IN THE EDUT: one field added to two verse files. With it
removed all 5,853 verse files are byte-identical to the previous L1.
L1 SHA-256 f743b34b…3f1494 → 9e1abe4e…3c3086
Moved together: L1Reader @l1_expected_hash · Luchot.Firmware + Spec ·
azazel goral._L1_INTEGRITY_HASH · the kohen's gate in KOHEN.md and
KOHEN_PROMPT.md. The goral's seed moves with it.
THE SMALL LETTERS. Not added, because there are none to read. The
builder was not dropping them: it reads `x-small` and records what it
finds. The source — the WLC as the Open Scriptures Hebrew Bible carries
it — has no small-letter mark anywhere in the five books. Not the aleph
of ויקרא, not the he of בהבראם. It marks four large letters (the vav of
גחון, the nun of משפטן, the ayin and dalet of the Shema), and L1 has
those four. The fuller Masoretic lists of enlarged and reduced letters
are a tradition of the scroll that this edition does not encode. To
engrave them would be to add to L1 something its source does not say;
that needs a second source and the transmitter's word, not a builder's
list from memory. l1/README.md, vayakhel §1.2 and terumah §E1 now state
this plainly.
(The puncta extraordinaria ARE in L1 — the upper dot U+05C4 is kept on
the text of 11 words.)
Doorpost @expected_hash b62311e7… → 3bc5dd6a…, in this commit, with the
witnessed file that moved it (vayakhel §1.2).
Read before committing — all five suites on the new L1: Elixir 496/0
(seeds 0 7 42) · Haskell 59/0 · korban 222 passed 1 skipped · azazel
125 passed · Prolog 6 suites exit 0. verify_l1.py OK. Four images
built; full local stack: the witness reads 9e1abe4e… and matches,
health 7/7, no errors; each image carries the new scroll read-only.
A trap met on the way: the first Elixir run after the Doorpost constant
was edited failed two Doorpost tests against the OLD constant. mix had
not recompiled doorpost.ex — the edit changed neither its size nor
(within the same second) its mtime. `touch` and re-run: 496/0. Images
compile from scratch and are not exposed to this.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
||
|
Scott KarpandClaude Fable 5.1
|
7d46fdc7c9 |
L1: the petuchot and setumot are engraved (tablet 39)
The scroll in the aron had no paragraphs. The WLC marks every petuchah
(פ) and setumah (ס); scripts/build_l1.py handled sof pasuq, maqaf and
paseq and read past these. L1 was 5,853 verses divided only by chapter
numbers, which are not Torah's.
Transmitter's word: "add the petuchah and setumah markers to L1."
- build_l1.py: a verse followed — or interrupted — by a break carries
"parashah_breaks": [{ "after_word": 13, "type": "petuchah" }]
The break is recorded by the word it follows; at the end of a verse
that is the verse's word count. text_hebrew is untouched: the break
is structure, not letters. Counts go into build_report.json.
- 293 petuchot and 394 setumot, in 686 verses — the source's count
exactly. Four fall inside a verse (pisqa b'emtza pasuq): Bereishit
35:22, Devarim 2:8, 5:21, 23:8; Devarim 5:21 has two. These are the
Leningrad Codex's divisions (a scroll written to the Rambam's list
has 290 and 379); the README says which tradition L1 records.
WHAT CHANGED IN THE EDUT: one field added to 686 verse files. Verified
file by file — with that field removed, all 5,853 verse files are
byte-identical to the previous L1.
L1 SHA-256 156b1d27…d8c9d5 → f743b34b…3f1494
Moved together, because they pin the hash: L1Reader @l1_expected_hash ·
Luchot.Firmware + Spec · azazel goral._L1_INTEGRITY_HASH · the kohen's
gate in KOHEN.md and KOHEN_PROMPT.md. The goral's seed changes with it;
azazel records already in L3 stand as cast.
- L1Reader: `parashah_breaks` on every verse (a list, empty where there
is none; only the two kinds are passed on).
- test/tablet39_parashah_breaks_test.exs — 7 tests: the six days of
creation each close with a petuchah; the Shema's first paragraph
(Devarim 6:4–9) opens after a petuchah and closes with a setumah; the
break inside Bereishit 35:22 falls between ישראל and ויהיו; Devarim
5:21 has its two; the whole scroll counts 293 / 394 / 686.
- Specs: vayakhel §1.2 and terumah §E1 said the marks were dropped;
they now say where they are. (The source element is <seg>, not a
milestone as
|
||
|
Scott KarpandClaude Fable 5.1
|
f4e8e5afc8 |
specs §1.1/§1.5 + Doorpost: the pins say what is built, and the build obeys them
The pin audit (2026-10-04) found Construction §1 wrong in three ways.
1. §1.5's mix.exs block was not the real mix.exs.
phoenix ~> 1.7.11 was never a dependency (the API is Plug on
Bandit); `ecto` is `ecto_sql`; telemetry is transitive (locked
1.4.2, outside the stated ~> 1.2.1); bandit, plug and the sanhedrin
path dependency were missing. The block now matches mix.exs and adds
the versions mix.lock actually builds. Terumah's Elixir table
matched the wrong block and also said postgrex ~0.17.4; corrected.
2. §1.1's component table described a machine that is not the one
running. PostgreSQL 16.2 via apt → 17, managed (17.11 today).
Podman 4.9.3 → Kubernetes/containerd, managed; Podman is used
nowhere. Ubuntu 24.04.1 → what each image is in fact built on
(Ubuntu 22.04 for pipeline and engine, Debian 12 for korban and
talmudic) and the managed node (24.04). rebar3 3.23.0 → 3.22.0, the
version in the build image. Azazel Processor added as a component.
The spec is corrected to the deployment, not the reverse.
3. §1.1's rule — "No >=. No +. No ranges." — was not true of the spec's
own Elixir deps, and "lock files are committed" was not true of two
components. Both halves fixed:
- The rule now says what is the case: a manifest may give a range;
the committed lock is the pin; every image is built from its
lock; base images are pinned by tag; managed services are pinned
to a major version and patched by the provider.
- And the build was made to obey it. The Korban image installed
three packages named by hand and let pip pick every transitive
one on the day of the build, plus an unpinned poetry-core. It now
installs the lock's main group by exact version AND hash
(scripts/lock_to_requirements.py; --require-hashes --no-deps;
pip check). The image holds the lock's 12 packages and nothing
else.
- azazel_processor/poetry.lock: generated (it had none).
- kushya_generator/pyproject.toml + poetry.lock: the four §1.4 pins
(llama-cpp-python 0.2.55, peft 0.9.0, transformers 4.38.2,
datasets 2.17.1) were pinned nowhere in the repo. They are now
declared and locked (74 packages). The lock resolves; it has NOT
been installed or run — Step 21 is unbuilt and needs a GPU.
Its test config gains the korban and azazel source paths, which
the external backend needs: 3 tests there had been failing since
the azazel build, unnoticed because that suite is in no total.
Doorpost @expected_hash 39b66c31… → 1cd07079…, in this commit, with the
witnessed file that moved it (vayakhel).
Read before committing: Elixir 474/0 (seeds 0, 42) · korban 222 passed
1 skipped and azazel 125 passed, each with dependencies installed from
its lock by hash · kushya 17 passed (was 14 of 17) · korban image
starts, imports korban_processor + azazel_processor, pip check clean ·
pipeline image reports specification_identity verified 1cd07079….
Left as written: architecture §2's technology table still names
Phoenix, PostgreSQL 16, Ubuntu 24.04 and rootless Podman for Sanhedrin
isolation, and terumah §B still lists systemd, squashfs, Caddy,
nftables for a self-hosted server. Those describe a design, not a pin.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
||
|
Scott KarpandClaude Fable 5.1
|
8e4f94c66f |
L1: the shoresh is read from the lexicon, not guessed (tablet 34)
L1's `shoresh` field was derived by stripping a prefix letter and a
suffix letter from the consonants and keeping the first three. וַיַּרְא
"and he saw" was engraved as ירא ("fear"); ויהי as יהי; שמים as שמ;
מזבח as מזב; הבאתי as בא. And parse_prefixes took a word's morphology
from the LAST morph segment, which for a suffixed word is the pronominal
suffix — so every suffixed noun and verb in Torah (לבבך, נפשך) was
classed as a suffix and written with no shoresh at all.
Every WLC word already carries its lemma, and the OSHB lexical index
files each lemma under its Brown-Driver-Briggs root (terumah §E2 named
BDB + Strong's as the root lexicon and marked it "needs consolidation").
scripts/build_l1.py now reads the root from that filing:
main the lemma is a root 23,750 words וירא → ראה
sub filed under a root 31,066 words מזבח → זבח
single stands alone in the lexicon 2,671 words משה → משה
none no etymology recorded → absent 3,002 words יוסף, מצרים
particles, pronouns → absent, as before 19,493 words את, אשר, כי
57,487 words carry a shoresh (was 51,836); 1,378 unique roots (was
2,608 fragments). Roots are written as the lexicon writes them — two
letters (בן), three, or more (שמאל).
WHAT CHANGED IN THE EDUT: only `shoresh`. Verified file by file — with
that one field removed, all 5,853 verse files are byte-identical to the
previous L1. Text, consonants, lemma, morph, te'amim, prefix, ketiv/
qere, large and small letters: untouched. The old builder, run against
the same morphhb checkout, reproduces the old hash exactly; the new
builder is deterministic (two runs, one hash).
L1 SHA-256 99bbd07a…4ea377 → 156b1d27…d8c9d5
sources morphhb 3d15126 · HebrewLexicon 21c9add
(recorded in l1/build_report.json)
Moved together, because they pin the hash or the count:
vav_pipeline L1Reader @l1_expected_hash · shoresh-engine Luchot.Firmware
(hash, uniqueRootCount) + Spec · azazel_processor goral._L1_INTEGRITY_HASH
· KOHEN.md / KOHEN_PROMPT.md gate · l1/README.md (new "Roots" section) ·
README.md (second quarry) · AGENTS.md · root_meanings.pl comment.
The Doorpost manifest is not touched; its hash does not move.
Note for the kohen: the goral's seed inherits the L1 hash, so lots cast
after this commit are cast from a different seed than the azazel
records already in L3. Those records are history and stand as written.
All five suites on the new L1: Elixir 461/0 (seeds 0 42 191243) ·
Haskell 55/0 · Prolog 6 suites exit 0 (swipl 9.2.2) · korban 222 passed
1 skipped · azazel 125 passed. Four images built; full local stack
(real engine binary, PostgreSQL 16 with SSL, migrations) answers
Bereishit 1:4 with ר-א-ה, health healthy, 6 interpretations in L3,
0 write failures.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
||
|
Scott KarpandClaude Opus 4.7
|
6eb555a5cf |
azazel_processor: commit 6 — protocol envelope (property 8)
Pass 4 / Finding F, commit 6 of 10 from BUILD_AZAZEL.md. Closes
the Python side of the build; commits 7-9 are cross-language
integrations (L3, observability endpoint, intake-side routing).
Vayikra 16: the Azazel goat goes ONCE A YEAR, on Yom Kippur, under
the full ritual envelope.
Vayikra 17:7: וְלֹא־יִזְבְּחוּ עוֹד אֶת־זִבְחֵיהֶם לַשְּׂעִירִם —
they shall no longer offer their sacrifices to the sa'irim. Same
destination outside the envelope is avodah zarah (the failure mode
specs/eish_zarah_reading.md catalogs).
THE GATE:
- is_yom_kippur_protocol_active(today=None) — predicate; True
when an operator-configured Gregorian window covers today
- authorize_disposal() — context manager; explicit per-thread
authorization for ONE disposal cycle (event-driven disposals
during the year)
- has_explicit_disposal_authorization() — predicate; True iff
the current thread is inside an authorize_disposal block
- assert_protocol_envelope_active() — gate; raises ProtocolViolation
if NEITHER condition is met. Wired into dispatch_for_disposal
as the FIRST check (before role gate, before any side-effect)
WHY a CONTEXT MANAGER, not a flag:
A persistent flag would let an operator "set authorization true and
leave it" — the avodah-zarah shape (using the channel without the
envelope). The context manager forces explicit scope: authorization
exists only within the with-block, only on the thread that opened it.
WHY PER-THREAD:
Process-wide authorization would let two concurrent disposals share
without explicit coordination — operator A's authorize_disposal()
would silently authorize operator B's dispatch_for_disposal(). Per-
thread isolation forces each disposal to carry its own authorization
explicitly.
WHY OPERATOR-CONFIGURED Gregorian DATES, not Hebrew calendar:
The Hebrew-calendar conversion lives in vav_pipeline (Elixir) — we
do not bring a Hebrew-calendar library into azazel_processor's
dependency surface. Operators set the Gregorian range each year (or
delegate it to a wrapper script). For now, this module is purely
data-shape; commit 9 may wire vav_pipeline's Hebrew calendar to set
the window automatically.
CREATED:
src/azazel_processor/protocol.py ~155 lines
ProtocolViolation — raised on out-of-envelope
invocation; error message
names BOTH legitimate paths
(Yom Kippur window, authorize
context) so the operator
knows their options
set_yom_kippur_window(start, end) — operator-configured;
inclusive Gregorian dates;
inverted range clears
clear_yom_kippur_window() — disable
is_yom_kippur_protocol_active(today=None)
— date-range membership
authorize_disposal() — context manager; reentrant
(nested entries each
increment a counter, each
exit decrements; cleared
via try/finally even on
inner exception)
has_explicit_disposal_authorization() — per-thread, scoped
assert_protocol_envelope_active() — the gate
CHANGED:
src/azazel_processor/azazel.py
dispatch_for_disposal now calls assert_protocol_envelope_active()
as the FIRST check. Outside the envelope: ProtocolViolation
raised before any side-effect (no shaliach activation, no payload
held, no checkpoint persisted).
test/test_azazel.py
Added an autouse `_envelope` fixture that wraps every existing
orchestrator test in `with authorize_disposal():`. Protocol
restriction is verified separately in test_protocol_restriction.py.
CREATED:
test/test_protocol_restriction.py 16 tests covering:
- default state (no window, no authorization) refuses dispatch
with a ProtocolViolation that names both legitimate paths
- assert_protocol_envelope_active raises in default state
- both predicates False in default state
- Yom Kippur window: today inside → active; outside → inactive;
inverted range clears; inclusive at boundaries; clear disables
- authorize_disposal: inside block → active; nested blocks
decrement correctly; exception inside still clears (try/finally);
per-thread isolation (sibling thread sees False even when parent
thread is inside an authorize_disposal block)
- dispatch proceeds inside authorize_disposal context
- dispatch proceeds inside Yom Kippur window
- outside envelope: NO side-effect (no in-flight file written, no
payload held by shaliach)
- protocol gate fires BEFORE role gate (inactive handler outside
envelope raises ProtocolViolation, not ShaliachError)
Per-test fixture clears Yom Kippur state on entry and exit so
module-level state doesn't bleed between tests.
PYTHON SIDE NOW FEATURE-COMPLETE:
Properties defended end-to-end via Python code:
2 goral
3 vidui
4 shaliach + tevilah
5 checkpoint persistence
7 authorized routing
8 protocol envelope
10 hard termination
11 classification discipline
12 unmanaged destination
Properties remaining (cross-language, in commits 7-9):
1 parity provisioning — commit 9 (Korban→Azazel intake)
6 out-of-band acknowledgment — commit 9 (chatzotzrot wiring)
9 observable disposal — commit 7 (L3) + commit 8 (endpoint)
Tests: azazel_processor 125 passed (was 109, +16). korban_processor
unchanged at 203 passed, 1 skipped. No regressions.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
|
||
|
Scott KarpandClaude Opus 4.7
|
c3267c59bf |
azazel_processor: commit 5 — orchestrator state machine (properties 5, 7, 10)
Pass 4 / Finding F, commit 5 of 10 from BUILD_AZAZEL.md. Largest
commit so far; pulls together vidui + shaliach + goral + stages and
adds the state machine + in-flight persistence.
THE STATE MACHINE (six checkpoints, persisted):
STAGED → AzazelStaged envelope tagged; payload held by shaliach
CONFESSED → vidui carried into the staged value
ROUTED → routing committed (always ERETZ_GEZERAH); authorized_by
+ shaliach attribution recorded
DESTROYED → payload references cleared from shaliach;
gc.collect() called to release; payload becomes
unreachable from any orchestrator-visible code path
ACKNOWLEDGED → out-of-band ack callback invoked (commit 9 wires
Chatzotzrot.signal(:moadim) here)
RECORDED → AzazelRecord persisted via callback (commit 7 wires
the cross-language L3 file-watcher bridge here)
After RECORDED completes, the in-flight checkpoint file is shredded —
the disposal is now permanent in L3, the in-flight artifact must not
linger.
PROPERTY COVERAGE:
Property 5 (provisioned infrastructure / sukkot):
Each checkpoint persists state to <in_flight_dir>/<case_ref>.json
via write-then-rename (atomic against crash mid-write). list_in_flight
returns partial disposals so resume tooling can pick them up.
The Mishkan analog: the sukkot along the wilderness route where
the ish itti can rest if the disposal halts.
Property 7 (authorized routing):
AzazelRecord.authorized_by carries the kohen_gadol identifier;
AzazelRecord.shaliach carries the ish itti identifier. The
routing destination is the ERETZ_GEZERAH sentinel (property 12).
The routing decision is OWNED and traceable; the destination
is not.
Property 10 (hard termination):
The DESTROYED checkpoint clears handler._payload_refs and calls
gc.collect(). The AzazelRecord contains the vidui — what was
disposed — but NOT the disposed payload itself. The to_storage_dict
serialization explicitly excludes payload-shaped fields.
CREATED:
src/azazel_processor/azazel_record.py ~95 lines
AzazelRecord — frozen dataclass; fields:
case_ref, vidui, routing, authorized_by, shaliach,
triggering_call, goral_outcome_id, disposed_at
.to_storage_dict() — JSON-safe serialization for L3 + in-flight
src/azazel_processor/azazel.py ~225 lines
AzazelInput — what is being disposed
dispatch_for_disposal(input, *, handler, confess_with,
authorized_by, in_flight_dir,
acknowledger, recorder, goral_outcome_id)
— the single legitimate
entry point; walks the
six checkpoints
list_in_flight(in_flight_dir) — surfaces partial disposals
DEFAULT_IN_FLIGHT_DIR — /tmp/shorashim/azazel/in_flight
test/test_azazel.py 21 tests covering:
HAPPY PATH:
- dispatch returns AzazelRecord
- record carries authorization + shaliach attribution
- record routes to ERETZ_GEZERAH
- record carries the full vidui
PROPERTY 10 CI INVARIANTS:
- record does not carry payload (by attribute name)
- serialized record does not contain the payload's marker string
- STRUCTURAL: AzazelRecord dataclass field set has no
payload-shaped names (catches future regression)
- STRUCTURAL: no recover/undispose/restore/rehydrate/retrieve
function in azazel_processor public surface (catches
future regression)
- payload references cleared from shaliach at DESTROYED
PROPERTY 5 CHECKPOINT TESTS:
- in-flight file shredded after RECORDED
- checkpoints persist incrementally (verified by inspecting the
file at ack/rec callback fire-time)
- list_in_flight returns partial disposals
- list_in_flight empty when no disposals
- persistence is atomic (file is valid JSON at every visit)
ROLE GATE (property 4 still enforced):
- dispatch refuses inactive handler
- dispatch refuses non-IshItti
FILENAME SAFETY:
- case_ref with slash → ValueError
- case_ref with NUL → ValueError
AzazelRecord VALIDATION:
- authorized_by required
- shaliach required
- to_storage_dict omits payload
- frozen (cannot mutate)
DEFAULTS:
acknowledger and recorder default to no-op callables. This is
deliberate for commit 5 — the state machine works end-to-end without
external side effects, and commits 7 (L3) and 9 (chatzotzrot) plug
the real handlers into the same callback slots. Tests can supply
their own to inspect mid-route state.
Tests: azazel_processor 109 passed (was 88, +21). korban_processor
unchanged at 203 passed, 1 skipped. No regressions.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
|
||
|
Scott KarpandClaude Opus 4.7
|
e6fee82dd7 |
azazel_processor: commit 4 — goral (property 2, blind routing)
Pass 4 / Finding F, commit 4 of 10 from BUILD_AZAZEL.md.
Vayikra 16:8 — וְנָתַן אַהֲרֹן עַל־שְׁנֵי הַשְּׂעִירִם גֹּרָלוֹת,
גּוֹרָל אֶחָד לַה׳ וְגוֹרָל אֶחָד לַעֲזָאזֵל.
And Aharon shall place lots on the two goats — one lot for HaShem,
one lot for Azazel.
Yoma 62a: a kalpi (shuffled vessel), drawn with both hands
simultaneously. The high priest cannot see the lots in advance.
He cannot shape the outcome.
THREAT MODEL:
An operator who, given the choice, would route preferentially —
assigning the "better" goat to processing and the "worse" to
disposal. The specification makes that preference structurally
impossible.
DESIGN (BUILD_AZAZEL.md §6, Q4):
cast_lot(input_id) is DETERMINISTIC per (input_id, today). The
seed is SHA-256 over (L1 integrity hash, today's UTC date,
input_id). The first bit of the digest is the outcome.
Properties guaranteed:
- Same (input_id, today) → SAME outcome (no shopping by re-roll)
- Different input_ids same day → outcomes can differ (~50/50)
- Same input_id across days → outcomes can differ (~50/50)
- Operators cannot pre-compute without invoking (SHA-256 over
L1 hash + date is not mentally computable)
- Seed inherits L1 tamper-resistance: rewriting the daily seed
requires rewriting L1, which has its own halt rules
The L1 hash mirrored from vav_pipeline/lib/vav_pipeline/l1_reader.ex
(@l1_expected_hash). Both must move together if L1 is rebuilt.
CREATED:
src/azazel_processor/goral.py ~155 lines
GoralOutcome enum LA_SHEM | LA_AZAZEL
(both legitimate; the goral
enforces blindness, not a
"good vs. bad" distinction)
cast_lot(input_id, *, today=None) deterministic per (id, day)
outcome
cast_lot_outcome_id(input_id, full SHA-256 hex digest;
*, today=None) used by L3 records (commit 7)
so audit can recompute
GoralError on malformed input
test/test_goral.py 17 tests covering:
- outcome shape (one of two; enum members distinct)
- same input same day → same outcome (anti-shopping; verified
across 50 re-invocations + 1000-iteration audit-shopping test)
- outcome_id is reproducible 64-char hex
- 100 different input_ids → 100 distinct outcome_ids
- same input across 30 days → 30 distinct outcome_ids
- distribution across 10,000 input_ids on one day → ~50/50
(±5% tolerance, well within SHA-256 binomial expectations)
- distribution across 365 days for one input → ~50/50 (±10%)
- input validation (empty, whitespace, None, non-string raise)
- default today = UTC date
- audit can recompute outcome_id from (input_id, today) — the
audit-trail half of property 2
NOTE: when the orchestrator (commit 5) decides WHEN to invoke the
goral. For inputs that are cleanly processable or cleanly
un-processable, the goral is NOT invoked — they go to their channel
deterministically. The goral is the tie-breaker for genuinely
borderline inputs (Q1 in BUILD_AZAZEL.md §6).
Tests: azazel_processor 88 passed (was 71, +17). korban_processor
unchanged at 203 passed, 1 skipped. No regressions.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
|
||
|
Scott KarpandClaude Opus 4.7
|
85155b23fb |
azazel_processor: commit 3 — shaliach + tevilah (property 4, role-separated operator)
Pass 4 / Finding F, commit 3 of 10 from BUILD_AZAZEL.md.
Vayikra 16:21 — וְשִׁלַּח בְּיַד־אִישׁ עִתִּי הַמִּדְבָּרָה.
And he shall send the goat into the wilderness by the hand of a
designated man (ish itti).
Vayikra 16:26 — וְהַמְשַׁלֵּחַ … יְכַבֵּס בְּגָדָיו וְרָחַץ אֶת־בְּשָׂרוֹ
בַּמַּיִם וְאַחֲרֵי־כֵן יָבוֹא אֶל־הַמַּחֲנֶה.
The sender washes his clothes and bathes his flesh (tevilah), and
AFTERWARD comes into the camp.
Yoma 67a: the ish itti is a dedicated operator role, prepared in
advance, distinct from the kohanim who handle sacred processing.
Contact with the disposal channel produces state that must be
reset before the operator rejoins the trusted boundary.
THE THREE ROLES IN SHORASHIM:
Owner — submits the offering (upstream caller)
Kohen — performs shechitah/zerikah/haktarah (korban_processor
stages.py enforces owner/kohen separation)
Shaliach — receives input the kohen REFUSED and disposes of it
through the Azazel protocol (THIS module)
Without tevilah, payload references would persist across disposals;
disposal-N would contaminate disposal-N+1 with information from N.
The Mishkan analog: the ish itti who returns from the wilderness
without tevilah brings desert-state into the camp.
CREATED:
src/azazel_processor/shaliach.py ~210 lines
IshItti class:
__init__ — sets _immutable_identifier (no other immutable
attrs; everything else must be cleared by tevilah)
activate() — begin a disposal; refuses if already active (no
state-mixing across disposals)
hold_payload() — register payload ref for later destruction
tag()/get_tag()— intermediate metadata during the disposal route
tevilah() — Vayikra 16:26: clears ALL non-_immutable_
attributes (discovered dynamically via vars(),
so future state additions are auto-cleared)
identifier — propagates onto AzazelRecords (commit 7) for
shaliach attribution
ShaliachError — raised on role-separation violations
assert_shaliach_active(handler) — runtime gate every disposal-channel
function calls at entry; the type-system gate is
AzazelStaged (property 11). Refuses non-IshItti
callers AND inactive handlers. Identity-checked
(isinstance), not duck-typed.
THE LOAD-BEARING CI INVARIANT (test_tevilah_clears_all_non_immutable_state):
Walks vars(handler) post-tevilah and asserts every non-_immutable_
attribute is in a 'fresh' state (empty container, False, None, 0).
A future contributor who adds `self._cache = {}` (or any state
attribute) without extending tevilah() to clear it breaks this and
the test fires before the regression ships. The discovery is
DYNAMIC — no hand-maintained list of attributes to clear; tevilah
is implemented to clear all non-_immutable_ attrs and the test
verifies it does.
test/test_shaliach.py 23 tests covering:
- construction (named identifier required, validates non-empty,
non-whitespace, string)
- identifier immutability across activate→tevilah→reactivate
- lifecycle: starts inactive; activate marks active; re-activate
without tevilah raises; tevilah→reactivate succeeds
- hold_payload/tag require active; both work when active
- tevilah clears payload refs, intermediate tags, sets inactive
- tevilah preserves _immutable_identifier
- THE CI INVARIANT (dynamic walk)
- assert_shaliach_active: passes for active IshItti, refuses
inactive, refuses non-IshItti, refuses duck-typed impostor
- integration: stale-state-across-disposals prevented end-to-end
NOTE ON PYTEST INVOCATION:
When running azazel_processor tests, use `--rootdir=.` from the
package directory. Without it, pytest auto-discovers /Users/scottkarp/
shorashim as rootdir and ignores the local pyproject.toml's
testpaths config. Documented here so future runs match.
cd azazel_processor
PYTHONPATH=src:../korban_processor/src python3 -m pytest --rootdir=.
Tests: azazel_processor 71 passed (was 48, +23). korban_processor
unchanged at 203 passed, 1 skipped. No regressions.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
|
||
|
Scott KarpandClaude Opus 4.7
|
a2744c012a |
azazel_processor: commit 2 — vidui (property 3, enumerative commit)
Pass 4 / Finding F, commit 2 of 10 from BUILD_AZAZEL.md.
Vayikra 16:21 — v'hitvadah alav et kol avonot b'nei Yisrael v'et kol
pish'eihem l'chol chatotam. The high priest confesses ALL iniquities,
ALL transgressions, for ALL sins. The verb is hitpa'el (reflexive) —
the confession is self-addressed before it is addressed to the
system. The kol-three-times structure is mandatory.
ENGINEERING READ:
The commit message for a disposal must enumerate what is being
disposed of. Untagged disposal is not disposal; it is loss. Azazel
does not accept uncommitted writes. The enumeration is produced at
the moment of commit, not retrieved from a pre-existing log.
WHY OPERATOR-PRODUCED, NOT AUTO-DERIVED:
The Mishkan high priest speaks the confession aloud. The system did
not generate it from a classification enum. In software: the caller
of confess() MUST supply the descriptions, not have them
auto-generated from a classification → string table. If the table
generated the confession, the operator could dispose state without
ever naming it themselves; the confession would become an
autogenerated label, not a self-addressed enumeration. That is the
failure mode property 3 defends against.
The convenience helpers (confess_empty_vital, confess_halt,
confess_boundary_violation, confess_malformed_input) DO NOT hide
this requirement. Each requires the operator to supply the specific
item; the helper only routes which category it goes into.
THE THREE CATEGORIES:
avonot — iniquities. CATEGORICAL errors: halt-trips
(avodah_zarah, gilui_arayot, shfichut_damim — the
cardinal sins). Sins of orientation.
peshaim — transgressions. BOUNDARY violations: distinction
collapses (gilui-shaped), L3 schema rejections.
Sins of boundary-keeping.
chatot — sins. TARGET errors: empty-vital, malformed responses.
The literal meaning of chet — missing the mark.
Most disposals fill exactly one category. The Vidui dataclass
REQUIRES all three fields (the enumeration is comprehensive at the
shape level), but the lists may be empty. What is forbidden is the
ABSENCE of a slot — and the case where ALL THREE slots are empty
(EmptyConfessionError, no untagged disposal).
CREATED:
src/azazel_processor/vidui.py ~250 lines
Vidui — frozen dataclass: avonot, peshaim, chatot,
confessor, confessed_at (all required;
categories are tuples for deep immutability;
__post_init__ rejects all-empty, empty-string
items, non-string items, anonymous confessor)
EmptyConfessionError — raised on untagged-disposal attempts
confess(*, avonot, peshaim, chatot, confessor)
— the operator-produced builder
Vidui.to_text() — kol-three-times rendering for L3 storage
Vidui.category_summary()
— counts only, no items (for the public
observability endpoint, commit 8 / property 9)
Convenience helpers (each requires operator-supplied item):
confess_empty_vital(korban_summary, *, confessor)
confess_halt(condition, detail, *, confessor)
confess_boundary_violation(violation, *, confessor)
confess_malformed_input(description, *, confessor)
test/test_vidui.py 27 tests covering:
- structural (exactly five fields, no silent additions)
- all category-fill combinations succeed
- all-empty / empty-string / whitespace / non-string items raise
- confessor required (TypeError) + non-empty/non-whitespace
validated (EmptyConfessionError)
- frozen dataclass (cannot mutate after construction)
- categories are tuples (deep immutability — frozen=True does
not propagate through container types)
- confessed_at: timezone-aware UTC, set to current time
- to_text() preserves kol-three-times structure; omits empty
categories
- category_summary() returns counts WITHOUT items
- all four convenience helpers route correctly + still require
confessor
Public surface updated in __init__.py.
Tests: azazel_processor 48 passed (was 21, +27). korban_processor
unchanged at 203 passed, 1 skipped. No regressions.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
|
||
|
Scott KarpandClaude Opus 4.7
|
d06ecd244d |
azazel_processor: commit 1 — skeleton + stages (properties 11 & 12)
Pass 4 / Finding F, commit 1 of 10 from BUILD_AZAZEL.md. Lays the
load-bearing foundation: classification discipline (property 11) at
the type system level, and unmanaged-destination opacity (property
12) via a Python sentinel. Every later commit in the Azazel build
inherits these defenses for free.
CREATED:
azazel_processor/
pyproject.toml — Poetry config matching
korban_processor; pythonpath
includes ../korban_processor/src
so the cross-package
classification-discipline test
can import both packages
src/azazel_processor/
__init__.py — public surface for commit 1:
AzazelStage, AzazelStaged,
AzazelStageError, ERETZ_GEZERAH,
is_eretz_gezerah
eretz_gezerah.py — the destination sentinel; module
__getattr__ refuses any
attribute beyond the two public
names so introspection cannot
discover further handles
stages.py — AzazelStage enum (six
checkpoints from the build doc:
STAGED → CONFESSED → ROUTED →
DESTROYED → ACKNOWLEDGED →
RECORDED), AzazelStaged frozen
dataclass, AzazelStageError
test/
test_eretz_gezerah.py — 11 tests covering: sentinel
identity stable across imports,
not-a-string, no public attrs,
repr leaks no destination info,
identity (is) not equality (==),
impostor with __eq__ rejected,
module __getattr__ refuses
unknown access, CI invariant
forbidding destination-handle
ASSIGNMENTS in source (catches
regressions; mention in
docstrings/__all__/getattr
allowlist is fine)
test_stages.py — 10 tests: intra-Azazel gating
(happy path, mismatch raises
naming both stages, frozen
dataclass, six distinct
checkpoints) and CROSS-PACKAGE
classification discipline:
KorbanStaged and AzazelStaged
are distinct types; enum value
spaces are disjoint; an Azazel
function refuses a KorbanStaged
input and vice versa; no
conversion function exists in
the public surface
PROPERTY COVERAGE THIS COMMIT:
Property 11 (classification discipline):
Defended by the type-distinctness of AzazelStaged vs.
korban_processor.stages.Staged. Cross-package tests verify the
two cannot be substituted at runtime; the absence of any
korban_to_azazel/to_korban/from_korban/convert function in the
public surface is verified by name-shape grep.
Property 12 (unmanaged destination):
Defended by the sentinel design: only object identity is exposed;
no string handle exists; module __getattr__ refuses introspection;
CI invariant catches assignment-shaped regressions across all of
azazel_processor/src/.
PENDING (commits 2-10 per BUILD_AZAZEL.md):
vidui (commit 2 — enumerative commit, property 3)
shaliach + tevilah (commit 3 — role-separated operator, property 4)
goral (commit 4 — blind routing, property 2)
azazel orchestrator + state machine (commit 5 — properties 5, 7, 10)
protocol envelope (commit 6 — protocol-restricted access, property 8)
L3 azazel_records (commit 7 — operator-internal half of property 9)
/peah/azazel_log endpoint (commit 8 — public half of property 9)
Korban→Azazel routing on empty-vital + chatzotzrot wiring
(commit 9 — properties 1, 6)
Closure pass (commit 10)
Tests: azazel_processor 21 passed. korban_processor unchanged at
203 passed, 1 skipped (live-API). No regressions across the boundary.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
|