Run the dsh-archive-agent-notes audit over every active Agent Note on
current master, judging each record by whether its rationale still guides
work rather than by size or age.
- Archive 453 implemented bilingual triplets (417,882 English words):
completed UI chrome, narrow adapters, closed bug fixes, implementation
walkthroughs whose package READMEs, docs pages, generators, or successor
notes now carry the useful behavior, and 51 records fully superseded by
a later active note. Keep 201 implemented notes whose ownership rules,
negative guarantees, durable or wire semantics, security rules,
reintroduction conditions, or still-tempting rejected alternatives
remain useful.
- Reject 7 proposals whose premise is gone or whose work shipped in
amended form under other records; delete 2 rejected notes that no
longer prevent a plausible mistake.
- Retarget every remaining inbound link to the archived path, and repair
active prose that named an archived record as the owner of a live fact:
parenthetical citations drop, ownership sentences redirect to the
README, docs page, or active note that states the fact, and history
citations say so. Chinese files link the English archived path because
the pairing gate treats the frozen tree as outside the bilingual corpus.
- Seal 1,359 new frozen artifacts; existing seals are unchanged and
outbound links from archived notes are neither inspected nor repaired.
- Regenerate docs/config-catalog.md after the hook-bridge comment edits
shifted two source line numbers.
The review's final wording items: the portable-identifier note's Scope said the
backend 'has since shipped' without noting it is experimental/private; the
RESERVED_WORDS JSDoc said backends 'ship for both languages'. Both now name the
TypeScript backend as released and the CPython backend as experimental and
private. The package README also records that the truncation-marker text and
tempdir prefix keep the pre-rename short names (byte-anchored by tests,
independent of the npm name).
The review's carry-over: 'each has a published backend' in the CodeRuntime
JSDoc and its projections (tool-cordis api-catalog, subsystems page) plus
'both shipped'/'backends ship' in the code-runtime README all claimed the
Python backend is released; it is private and experimental, excluded from the
release family. The wording now states the TypeScript backend is released and
the Python backend is experimental and private (not published), in the JSDoc
(api-catalog regenerated to match), the READMEs (paired), and the subsystems
page (paired).
The review's move-follow-ups: the Windows test exclude now points at
packages/experimental/code-runtime-python (the constructor throws by design on
Windows, so the suite must stay excluded); the invariant companion and
@module annotations use the new npm name; the truncation marker text and
tmpdir prefix stay as-is (tests anchor them); the package JSDoc and READMEs no
longer call the private experimental backend 'published'/'shipped'; the
code-runtime README row describes the package as protocol AND runtime; the
fd-3 note records the package's experimental location.
The CPython code runtime's complete public contract is experimental, so it
moves to packages/experimental per the experimental-packages rules: npm name
@deepseek-ai/dsh-experimental-code-runtime-python, private: true, no
publishConfig. All references updated (code-runtime READMEs, config-catalog
and module-graph regenerated with zh alignment, tsconfig paths, doc-standard
and workspace-constraints scripts, the fd-3 and settlement Agent Notes, and
the package README links); md-links and translation pairing pass, and the
suite still runs green.
The reviewer's standing item: the unknown-binding reply ran JSON.stringify on
the WHOLE capped target (global + '.' + name, each up to maxValueBytes code
units), allocating the escaped form — up to ~6x under control-heavy input, a
multi-hundred-MB spike near the maxValueBytes ceiling that no hostile-peer
bound would have admitted. The escaped preview is now built from a 1 KiB
prefix of the target (enough to identify the binding); capMessage still
enforces the reply budget. A forged huge-name case drives the path.
The review's wording item: the layer-5 bullet ended with 'not by this PR'
(zh: 'not borne by this PR'), which references PR context in durable prose.
The sentence now ends with the current-state fact ('not by this package's
suite'), paired and re-recorded.
The review's open suggestion: the Known Limitations now records that the
real-Loader assembly snapshot is deferred to issue #1182 layer 5 (this package
is exercised through ctx.plugin and real-subprocess tests; the full application
composition is covered by a tracked assembly test in that layer), paired.
The review's two remaining non-blocking items: (1) a case where the run ends
with a SEALED open hold (past MAX_PENDING_CHUNKS) — finish() must commit the
sealed prefix, verified to fail if finish drops openSealed. (2) openSealed is
now a block ARRAY (one joined block per seal) matching the fd-3 reader's
blocks and the stray capture's seal, instead of one repeated string concat
that leaned on V8 ConsString amortization.
The review's follow-ups on the open-seal fix: (1) the settlement note's seal
section now records the HOST-side open hold seal (openParts -> openSealed,
mirroring the child _LogStream and stray-capture seals), paired. (2) the
first-fragment guard comment notes the empty-first-frame case (bills cost + 1 =
3, establishes no hold, bounded over-charge in the safe direction). (3) a
regression case commits a SEALED open hold before the truncation marker —
verified to fail if truncateLogs drops openSealed.
The review's warning: each held open fragment is a distinct array slot plus
string object header (~30x overhead the byte cap cannot see), and a
budget-sized single-character open flood is honest-child reachable
(print('x', end='', flush=True) in a loop). With maxLogBytes near its ~67 MB
load ceiling that was up to ~2 GB of host auxiliary heap. The hold now seals
into one block past MAX_PENDING_CHUNKS, mirroring the fd-3 reader's blocks and
the stray capture's seal; the merge, truncateLogs, and the finish residual all
read sealed + current fragments, and a within-budget flood regression asserts
the merged entry is byte-identical.
The review's wording items: the load-check comment still referenced the
resolvePythonBin JSDoc's old ENOENT promise; the spawn-site comment called the
type assertion a non-null assertion; and two comments claimed the
'logs serialize to maxLogBytes + marker + envelope' bound is recorded in the
README's Known Limitations, which has no such entry — the cross-references are
dropped, the bound stays stated inline.
The review's carry-over: detachResidual (a test seam for the settled run's
resource cleanup) is re-exported from the '.' entry but was not in the README's
declared public surface; the list now names it alongside resolvePythonBin and
readProcessStart, paired.
The empty-open continuation skip (5b61a8fe6) made the Known Limitations entry
stale — the held fragment array no longer grows per empty frame — so the entry
is removed on both sides. The public-surface list now declares
resolvePythonBin and readProcessStart, which the '.' entry re-exports for the
test suite.
The skip branch (an empty open continuation is not pushed into the hold) needs
coverage; a case drives an empty continuation between a first fragment and the
closing frame and asserts the merged entry is unchanged.
The review's items: a zero-content open continuation bills 0 but still pushed
'' into the held fragment array, so a forged empty-open flood grew host memory
without touching the ledger — the push is now skipped (an empty fragment
contributes nothing to the merged entry). The spawn-site comment said a PATH
change between load and run would fail with ENOENT; it actually makes spawn
throw synchronously, which the surrounding try settles as worker-exit.
The review's warning: the pythonBin load-rejection is a product-visible change
(unresolvable basename now fails at load instead of a run-time worker-exit),
but the READMEs (en + zh) only said the basename is resolved against PATH, and
the load-check comment still described the old fallback. The README pythonBin
entries and the load-check comment now state the rejection; pairing
re-recorded.
The review's follow-ups: (1) the product-visible change (an unresolvable
basename pythonBin now fails at load instead of a run-time ENOENT worker-exit)
is registered in the settlement note, paired. (2) PYABS falls back to the bare
name when python3 is not resolvable, instead of interpolating the literal
'undefined' into the wrappers.
The review's follow-ups on the pythonBin change: (1) the JSDoc and the two
call-site comments still described the old fallback-to-bare-name contract;
they now state the load-rejection behavior. (2) the ExceptionGroup case's
version guard raised a skip message on Python < 3.11 but the assertion still
required the truncation marker unconditionally — the assertion now matches
either the truncation marker (3.11+) or the skip message (3.10). (3) the shell
wrappers quote the resolved interpreter path.
The review's two non-blocking items: (1) resolvePythonBin returned the bare
basename when PATH had no hit, and spawn (env:{}) would silently fall to
execvp's platform default PATH and could start a system interpreter the caller
never asked for. It now returns undefined for an unresolvable basename and the
load check rejects it (absolute paths pass through), so the failure is loud at
configuration time instead of silent at spawn; the case that expected a
run-time worker-exit now asserts the load rejection, consistent with the
empty/NUL pythonBin cases. (2) the over-cap exception-group case skipped on
Python < 3.11 (ExceptionGroup is a 3.11+ builtin), matching the TaskGroup
case's version guard.
The review's portability warning: the six shell wrappers exec'd a bare
'python3', which /bin/sh resolves against its compiled-in default PATH while
the runtime spawns with env:{} — in environments where python3 is reachable
only through the caller's PATH (Nix, pyenv) every wrapper run would fail as
worker-exit. The wrappers now bake the resolved absolute interpreter path
(module-level resolvePythonBin, which the product spawn already uses), and
resolvePythonBin is exported for the tests.
The reviewer's residual timing item: the inherited-SIGXCPU reset ran AFTER
setrlimit(RLIMIT_CPU) and the boot-namespace construction, so a huge namespace
under an inherited ignore/block could burn past the soft limit inside that
window and be misclassified as worker-exit. The reset now happens at the very
top of _run, before the resource-limit setup and namespace construction.
The review's two follow-ups on the inherited-SIGXCPU fix: (1) a discriminating
case — pythonBin points at a wrapper that ignores SIGXCPU before exec'ing
python3, so the child genuinely inherits the ignore; with cpuSeconds: 1 the
busy loop must end as timeout (the bootstrap reset restored SIG_DFL), and
reverting the reset leaves it running to the wall — verified red. (2) The zh
README's OUTER wire section now carries the truncation-exception sentence
(the previous commit had duplicated it in the inner section instead); the
duplicate is removed, and the settlement note registers the inherited-SIGXCPU
reset.
The reviewer's standing issue: the child inherits the host's SIGXCPU
disposition and signal mask — if the host ignores or blocks SIGXCPU, the soft
RLIMIT_CPU fires but cannot stop the child, and the hard limit's SIGKILL then
classifies a definite CPU overrun as worker-exit instead of a timeout. The
bootstrap now resets SIGXCPU to SIG_DFL and unblocks it before any model code
runs (the settle-time enforcer already restores SIG_DFL for a program that
traps or masks the signal mid-run; this closes the inherited-state gap). The
zh README's outer wire section also gains the truncation-exception sentence to
match the en side.
The review's warning: the READMEs (outer and inner wire sections, en + zh) and
the fd-3 protocol note still claimed the next log frame always merges into an
open entry, while truncateLogs commits the already-billed prefix as its own
entry before the marker. The one exception (truncation) is now stated in both
READMEs and the owning note, paired and re-recorded. The prefix-commit case's
parenthetical describing the pre-fix implementation is removed per the
comment-does-not-record-review-history rule.
The reviewer's standing issue: line.toString('utf8') silently replaces illegal
bytes with U+FFFD, so a forged frame could land a corrupted completion value
(the honest child's lossless encoder never emits non-UTF-8, so such a frame is
hostile traffic). The fd-3 frame decode now uses a fatal UTF-8 decoder: an
illegal byte throws and the frame is dropped, same treatment as the
unsafe-integer check. A forged illegal-UTF-8 done frame is verified to be
dropped (the run settles on the program's real return), and reverting to
toString makes the case fail.
The review's warning: a flushed unterminated line is billed and committed
(README wire contract says so), but every truncation arm — the child truncated
frame, an over-budget open frame, an over-budget closing frame, and admit's two
budget arms — pushed only the marker, dropping the held prefix: the ledger
charged for output that vanished. All arms now funnel through truncateLogs(),
which pushes the (already billed) held prefix before the marker and clears
openParts, so the prefix survives and only the marker stays last; the finish()
guard drops the now-dead !logsTruncated check (a truncated run has an empty
hold). A regression case asserts [prefix, marker]; the forged-flood and
closing-overflow cases now expect the committed prefix plus the marker.
The review's suggestion: an empty open continuation frame bills zero and holds
one host slot, so a forged empty-open flood grows the held fragment array
without touching logBudget. Accepted as a residual (per-frame host cost far
below its ~30-byte fd-3 wire cost, bounded by pipe throughput, model-code trust
level equal to bash) and now registered in the README's Known Limitations on
both sides, paired and re-recorded.
The review's revision: the buffered-chunks pre-check's open-aware overhead is
observationally inert — when the +3 form trips and the open-aware form does not
(pending + newline in [remaining - 2, remaining]), _push_bounded_prefix
re-slices the same newline-free line text and _push_locked admits it under the
same open-aware billing, byte for byte. The comment now states that the
open-aware form keeps _push_bounded_prefix's 'certain to reject' precondition
true, contrasting with the scan pre-check whose slice carries the newline and
therefore genuinely truncates.
The review's warning: the en README's inner 'Wire contract' section still
described only the truncated flag while the zh counterpart (and the outer 'The
wire' section) described open. The inner en section now matches. The exact-fit
closing-line case comment described the buffered-chunks pre-check recipe while
the program actually drives the scan pre-check; the comment now states the
actual arithmetic and path (and the buffered variant was dropped — its writes
coalesce into one call in the test environment, so it did not discriminate).
The review's warning: while an open entry accumulates, the newline pre-checks in
the write path still charged a NEW entry's +3 cheap-bound overhead (quotes +
separator), so an exact-fit merged TAIL was truncated (or the pre-check
over-rejected it and flushed a truncated prefix). Both pre-checks now charge
the overhead only when no open entry is in progress, matching _push_locked's
open-aware bound. A regression case (the review's recipe: flush an open
fragment, then write one exact-fit newline-terminated line) is verified to
truncate when the +3 is restored.
The zh README's wire-contract section now describes the open flag like the en
side (the fd-3 Agent Note holds the split-billing arithmetic; a cross-doc link
was omitted to keep the bilingual link sequence aligned).
The review's warning: the one-byte overflow case ran through the CHILD ledger
(print path), so the host's first-fragment cap (logBudget - 1) never executed,
and the sub-2-byte guard test does not discriminate logBudget from
logBudget - 1 (a reverted cap still trips the guard). The frame is now forged
on fd 3, so a reverted cap of logBudget admits it and flushes it at settlement
— verified to turn the test red.
The review's dedupe suggestion: the split-billing arithmetic was stated in both
notes; the settlement note's Decision paragraph now links to the fd-3 protocol
note's wire-contract section (one home per fact), paired and re-recorded.
The new jsonStringCostUpTo guard (returns undefined below a 2-byte cap) was
uncovered: forged open frames drive the host ledger down to one byte, and a new
open entry's first-fragment cap (logBudget - 1 = 0) trips the guard and
truncates to the marker, asserted as the merged entry plus the marker.
The review's arithmetic checks: the closing-frame walk used cap
logBudget - openCost, so a compliant merged entry (58-byte wire cost under a
64-byte budget) could see a negative cap and truncate; the first-fragment cap
used logBudget instead of the ledger's logBudget - 1, so an open frame costing
63 was admitted with a bill of 64, pushing the ledger negative and letting a
subsequent empty frame ride in one byte past the configured cap; and the child
billed a closing frame as a fresh entry (quotes+separator again) instead of the
merged tail, truncating an exact-fit 30+30 entry.
Fixes: first-fragment cap logBudget - 1 (matching admit), continuation and
closing-frame cap logBudget + 2 (billed without quotes), jsonStringCostUpTo
returns undefined below 2 bytes, and the child's split billing keys off
_open_started alone (a closing frame pays content only) with the cheaper bound
len(text) while a merge is open. Regression cases cover all three arithmetic
paths.
The review's critical: the open-merge branch re-joined and re-walked the whole
held text per frame, so k tiny open frames cost O(k * budget) (thousands of
1-byte frames against a near-64 MiB budget would re-traverse hundreds of GB and
block the host event loop). The host now holds a fragment ARRAY with an
incrementally billed cost — each fragment's jsonStringCostUpTo walks only its
own text — and the closing frame bills only its own content, so the merged
entry's wire cost is charged exactly once, split across the fragments. The
child bills symmetrically: the first open fragment pays quotes+separator, each
continuation pays only its content, matching the host ledger (the review's
warning: per-fragment full billing truncated a 16-char merged entry under
maxLogBytes: 64 that costs only 19 bytes as one entry).
Regression cases: 16 single-character flushes merge to one whole entry; a
closing frame that overflows the remaining budget truncates to the marker; a
closing frame after an open flood already truncated the ledger is a no-op; and
a forged open-frame flood stays bounded by the ledger. The closing-frame
post-truncation guard is an invariant-false branch (an open frame that would
trip the ledger resets openParts, so a non-empty hold implies no truncation)
and carries a v8 ignore with that reason.
The review's critical: the open-merge branch accumulated the held fragment
before any ledger check, so a forged open flood could grow host memory without
touching logBudget. The held fragment is now bounded by the exact-cost walk
(jsonStringCostUpTo against the remaining budget; the closing frame's admit()
still bills the merged entry once), and the open field is registered in the
README wire-contract section and the fd-3 protocol note (en + zh). A forged
open-flood case asserts truncation to the marker under a 64-byte budget.
The review's remaining warning: an explicit flush of an unterminated line
(print(..., end='', flush=True)) pushed a full log frame, so the following
print() landed in a second entry and logs.join('\n') rendered 'a\nb' for what
the program printed as one line — a model-visible output defect. The flush
frame now carries an flag (LogMessage gains the optional field on both
sides and in the mirror test), the host holds it and appends the next log frame
to the same entry, and finish() admits the residual if the run ends with it
still open. The settlement note registers the decimal-context fix from the
previous commit.
The review's critical: Decimal(repr(value)).normalize() read the process-global
decimal context, so a legitimate program setting getcontext().prec = 2 silently
rounded the completion value's digits and traps[Inexact] = True made the encode
raise, misclassifying a successful run as an exception. A fixed module-level
Context(prec=28) makes the spelling decision context-independent; a regression
case mutates both context knobs and asserts the float round-trips exactly.
The binding-reply README entry now states the fact (no seam-level cap;
maxValueBytes meters only the done frame; a wide reply is rebuilt and encoded
whole, bounded by process memory), matching the earlier reviewer wording.
The review showed the added dispose case was a placebo (dispose in the same
tick as run means SIGTERM hits the group before the program body runs; the
group-emptied arm is already deterministically covered by the same-group
survivor case, which this removes the v8 ignore for). The test is deleted; the
stale silently-discards comment in the boundary test now says rejects; the
README Known Limitations gains the late-log-frame-drop and host-side
binding-value-memory entries. Pairing re-recorded.
Turtle