Three findings from review survived against the staged-save head:
The search provider read its options thunk per property, so a settings write
landing inside credential resolution sent the key resolved from the old
section to the endpoint named by the new one. Each operation now snapshots
once at its entry and threads that snapshot into credential resolution; a
regression test drives a commit into the middle of a search and pins that the
endpoint, model, and key all come from the section the search started on.
The /client entry exported components, controllers, and namespace constants
with no consumer, which the client export discipline allows only with sign-off.
Only types remain. The duplicate per-card Injected/Face interface pairs are
one declaration each now, so a member added to one side cannot silently miss
the other.
The credential state carries the reference it describes and its writability: a
reference change no longer projects the old answer onto the new name, an
out-of-order response for a stale reference is dropped, and a key that a
deployment sources from the process environment disables the control instead
of inviting a write the Host must refuse.
Also corrected three prose claims against the code they describe: the card's
fields do not differ by platform (the served schema does), the section's empty
line counts registered rather than visible cards and is read once, and the
search README overstated what a configuration surface learns about a key.
Master moved every workspace edge to workspace:^ and added release-member
fields; this branch's manifests follow, keeping only the dependency edges it
contributes. The client-runtime README keeps this branch's paragraph: master
did not touch it, and the base/user layers and `unset` it describes are what
this branch added.
release:vendor --prerelease rc.1 appends the identifier to the version each
package would take, so a first publication can be thrown away instead of holding
the stable dist-tag: publish already routes any version with a prerelease
segment to --tag next.
A prerelease does not consume its release numbers, so the stable release that
follows takes the same ones rather than skipping a patch. Deciding that needs
semver precedence, which git's version sort does not provide — it places
4.0.1-rc.1 above 4.0.1 — so the newest published version is chosen by comparing
versions here, numeric prerelease fields numerically.
dsh-sandbox-local declares @deepseek-ai/node-addon-landlock-run in
dependencies, not optionalDependencies, so omitting optional dependencies left
npm resolving it from a registry that does not carry it. The dsh pack job now
packs that entry for verification; its own platform packages stay out, being
optional and needing a musl toolchain per architecture.
The verification reads each directory by its contents rather than a pack order
file, because a directory packed only to satisfy a cross-sequence dependency has
no release order to describe.
The Landlock platform packages sit behind optionalDependencies, and npm fails
the install on their 404 rather than skipping them: they belong to the native
sequence, whose pack needs a musl toolchain and one build per architecture, so
this job cannot produce them and holds no credentials to fetch them. A consumer
that cannot install them must still start, which is what optional means here.
The release spec also gains block bodies where the lint rule rejects returning a
void expression from an arrow shorthand.
The root manifest carries the dsh family version. bump writes it with the
members, because the workspace constraint requires them to match, and that
constraint now accepts a prerelease segment: without both, release:dsh 0.0.2
left the root behind and 0.0.1-rc.1 could satisfy neither check.
The Landlock workflow no longer passes --access public, which overrode the
restricted publishConfig this repository just adopted for those packages.
Vendored change detection reads build inputs when a package publishes build
output, and vendor/cordis publishes the src its export map already pointed at:
its lib/ is untracked, so a real source edit read as 'nothing changed' and the
next publish would fail on a version whose bytes moved. The next version also
takes the last published version as its baseline, so a re-sync that restores a
lower upstream version cannot recompute a version already on the registry, and
bump confirms the registry carries what the newest tag names.
Tag prefixes are constructed rather than recovered from a full tag, which a
hyphenated version defeated. Pack runs group per ref so concurrent pull requests
stop displacing each other, the publish job carries the global group, and the
unused id-token permission is gone.
Every release script sits behind an entry guard, which is what lets the pure
judgements carry tests: tag naming, publish order and cycle reporting, version
arithmetic, payload policy, and the change judgement.
The Agent Note moves to implemented and states what shipped: one probe command,
the registry confirmation that now exists, and byte reproducibility recorded as
assumed rather than measured.
The invariant companion rule pinned @deepseek-ai/dsh-invariants to a ^0.0.1
peerDependency, which the workspace protocol replaces; it now requires
workspace:^ like every other workspace-internal reference.
The release note stated the provenance risk by naming the internal repository,
which the public-link gate rejects in tracked files. It now states the same
constraint without naming it.
Also corrects the Chinese text where decisions moved after it was written: the
release set is every member of packages/, apps/, and vendor/ rather than a
dependency closure, because plugins are mounted by name from cordis.yml and a
closure misses runtime-required packages; the Landlock packages publish
privately with the rest; and the manifests carry repository metadata alongside
the access level.
packages/, vendor/, and native/ carry different version baselines and change
at different times, so each releases on its own bump sequence, tag prefix, and
workflow. Records the vendor change judgement (per-package tags as the
last-published pointer), the registry-state publish rule that makes re-runs
idempotent, and the workspace:^ prerequisite for publishing at any version
other than the current 0.0.1.
English counterpart and pairing record follow once the Chinese text is
reviewed.
The web-app bundle gained dsh-app-boot on master beside this branch's
dependency list; both rows stand, and verify-cordis-config confirms every
patch row still resolves from that manifest.
Merging master's rename of the client manifest field left this package on the
old `dshClient` name. The row still composed and its empty node half still
activated, but the browser roster scan never matched it, so the whole settings
section vanished with no error anywhere.
verify-cordis-config now requires a packages/client package's "./client"
export and its dsh.client declaration to agree in both directions; the
composition file cannot tell a surface plugin from a Host plugin, so the
manifests are where this is checkable. The check is scoped to that group
because a Host package's "./client" export is the typed wire face its browser
consumers import, not a plugin the roster serves.
Replaces the patch round trip. An app's entrypoint resolves the command
line into a service, and the rows it configures read that service from
their own config — port: !!js ctx.get('webStartup')?.port ?? 3080 — so the
resolved value beats the value written beside it and nothing is written
back into a row or handed to the launcher.
A bundle names the entrypoint row in its manifest (dsh.bundle.entrypoint),
which is what lets the boot mount in two passes: entrypoints alone, then
the whole composition. That ordering is required, not cosmetic — a row's
config expressions are evaluated when the include applies the row, and a
strict ctx.get only answers for a service whose providing fiber is already
active.
What this removes: ctx.appPatches and the launcher-owned patch layer, the
disable/re-enable recycle and its in-flight-mount barrier, overrideConfig,
and the reload hazard they existed for. A live config edit now re-applies
the second pass against services that are still up, so a served port
survives by construction.
What it adds: ctx.appReady, because Loader settlement no longer means the
app is up — a row mounted in the second pass can observe a settled tree
while that pass is still running, or already rolling back. The web URL line
waits for it, so a boot that fails in the second pass announces nothing.
The Agent Note keeps the three vendored-Loader facts the mechanism turns on
— a row's config is resolved and validated when its fiber is created, while
it is still waiting; an inject update loses the plugin's static injections;
a row cannot be inserted from inside a mounting plugin — with the
alternatives they ruled out.
Resolves the vendored-package rescope (cordis -> @deepseek-ai/cordis,
schemastery -> @deepseek-ai/schemastery) against this branch's settings
namespaces: the bash capability namespace, its two executors' section
installs, and the new plugin-config client package all move to the scoped
specifiers.
Controls committed on blur, which turned leaving a field into a durable,
revision-fenced document write the user could neither preview nor undo, and
silently discarded a draft the field did not accept.
A card's form now owns the staged text every control renders, and Save is the
only point where drafts become writes. Reset stages the composed default the
same way; an invalid draft blocks the save with its reason instead of being
dropped; Discard drops the drafts; a collapsed card marks that it holds some.
The Host stays the only authority on whether a value was accepted, so the save
reads the section back and keeps the drafts of a save that did not land.
Every harness package declares cordis as a peer dependency, so publishing the
harness publishes the vendored framework layer too; under the upstream names
that publication would squat them on the registry.
scripts/rescope-vendor.ts owns the rename: the nine-package mapping, a
delimited-token rule that leaves cordis.yml, the Loader's cordis: builtins and
vendor directory names alone, per-file exemptions where a name is a directory
or an upstream runtime identifier, and the exact edits for sites a token rule
cannot express — dot-notation lookups, unquoted manifest keys, a regex literal
whose failure would make every Context-merge scan silently find nothing, the
vendored-manifest table, and the contracts that told readers vendored packages
keep their upstream names.
Markdown follows the rename inside every fence, because a fence is code a
reader copies or configuration they mount, and in `docs/` prose as well, where
a sentence quoting a name teaches something this repository no longer resolves.
Prose elsewhere records what was true when it was written, and the same
spelling can mean something else: the Python SDK's `cordis` option, or the
unvendored `@cordisjs/plugin-http`. `docs/rescope.md` states both names on
purpose and is exempt.
exactEditState() classifies every exact edit as pending, applied, or invalid.
An insertion keeps its anchor and a deletion keeps its remainder, so each side
counts the form that survives: a duplicated insertion, a half-applied
replacement, and a deletion whose remainder moved are all invalid. The run
classifies every edit before writing anything and aborts on the first invalid
one, so a disagreement between the mapping and the tree cannot leave a
half-rescoped checkout; each write re-reads its file, because two edits can
target one. rescope-vendor.spec.ts pins those rejections, and --check asserts
the whole post-state from the hygiene gate, so CI owns the invariant.
--reverse restores the upstream names, verified as a round trip: reverse, then
apply, reproduces this tree byte for byte.
docs/rescope.md is the consumer-facing reference: the old-name/new-name table
with each package's role, what the rename deliberately leaves alone, the sites
callers must change, and the commands to apply, verify, and revert. The Agent
Note carries the decision and its consequences.
The rename itself lands in the next commit, produced by running the script.
