DSH 0.1.5 owns the right column (its native Sidebar hosts every tab type
this plugin registers), so the plugin's own right panel, its width drag /
new-session seed prefs, and the free-window feature are gone. The bottom
workbench is the only plugin-owned surface left, and its expand/collapse
control is registered into DSH's session-header utilities list slot (the
header's corner seat belongs to the native sidebar's own control).
- state: one `bottomSplits` tree; `makeDefaultState()` takes no arguments;
`toggleBottomPanel` / `openTabInBottomPane` are the only landing
reducers; `setBottomHeight` and `sanitizeState` share the
`CONVERSATION_MIN` cap (a persisted tall height no longer snaps on the
first drag).
- service: every landing in the workbench (creation or focus) expands it;
`SIDEBAR_FEATURES` drops `'floatWindows'`; the open/close/activate tab
pools cover the single tree.
- shell: `Sidebar.tsx` renders only the bottom workbench; the toggle lives
in `sidebar/bottom-toggle.tsx`; layout-push pushes height only (no width,
no `--dsh-sidebar-width`).
- prefs: `openByDefault` / `defaultWidthPercent` / `changesDiffFloat`
removed from the schema, defaults, client parser, settings page, and all
20 dictionaries; the retired keys in stored documents are ignored.
- tests: free-window specs deleted, panel/float assertions dropped,
remaining suites adapted (124 files / 1287 tests); the mount lane asserts
the header toggle and sweeps the native sidebar's guide page.
- docs: README / README_EN / guide §0,§11 / AGENTS test map.
The guide renders `entry.icon`, and the `files` kind takeover registered
without one — so after the editor type stopped contributing its own
duplicate row (b56ceec), the single "Files" row was the only entry in the
new-tab list with a blank icon slot. The takeover now carries the editor
descriptor's icon (the takeover IS that descriptor's page).
Extracted `guideIconOf` so both registration sites build the icon field
the same way; the lifecycle test asserts the takeover's guide entry has
an icon.
The editor type's page identity IS the `files` kind takeover (same
explorer, same title), so the guide/new-tab list showed two identical
"Files" entries — observed on a real profile after the takeover landed.
The editor type now contributes no guide entry of its own; it stays
registered as the file RESOURCE viewer. Covered by the registration
lifecycle test (editor: no guide, files takeover: guide).
The editor type serves both the files PAGE and every file RESOURCE; its
native `title` callback ignored the address, so every opened file showed
up in the strip as "Files" — indistinguishable tabs. Verified on a real
profile: after opening `AGENTS.md` from the tree and a chat file link,
the strip read ["文件", "文件", "文件"].
`title(address)` now returns the file name for a file address and keeps
the descriptor title for a page address (`sidebar://editor`). Covered by
the registration-lifecycle test in tests/native-surface.spec.ts.
The native seat declares `sidebar.right.pane.tab` BEFORE it provides
`sidebarRightTabs`, so driving the registration off the slot declaration
read the service as missing and registered nothing — permanently, since a
declaration never collapses. Observed on a real DSH profile (0.1.5-alpha.1,
`dsh web` on the web profile): the plugin host and its other slot
registrations mounted, the native guide page stayed empty, and a probe
printed `sidebarRightTabs= undefined` at declaration time with the registry
present three seconds later. The scratch mount lane never saw it because
there the plugin activates after the native package finished providing.
`registerNativeSurface` now waits on the SERVICE (`ctx.inject`), which
re-runs whenever the registry appears/reappears, and keeps the per-body
slot registrations on `slots.inject` (they only need the declaration).
Regression test: `tests/native-surface.spec.ts` models the real order
(slot callback fires first, registry arrives later) and fails on the old
implementation. Verified on the real web profile: all guide entries
(files / editor / git / subagent / sidechat / terminal / browser) appear
and the terminal tab renders.
TabDescriptor requires `component`; the fixture omitted it, so CI's
typecheck step failed on the phase-2 branch (vitest never compiled the
file's props strictly, so the local run stayed green).
Phase 2 of the 0.1.5 adaptation: the plugin stops drawing its own right
panel and contributes tab TYPES to DSH's native right Sidebar, which owns
the right column. The bottom workbench (terminal dock) stays plugin-owned.
Registrations (src/client/native/):
- every TabDescriptor becomes a native tab type (kind = descriptor.id, the
default `extension` band) plus a keyed body in `sidebar.right.pane.tab`
and a live chip title in `sidebar.right.pane.tab.title`
- the `editor` type also claims `dsh-resource://file/**`, so it outranks
the built-in `text` preview (fallback) and every file the product opens
lands in the plugin's editor
- the built-in `files` page kind is TAKEN OVER by an extension
registration of the same kind: `openTab('files')` draws the plugin's
explorer, and the built-in resumes when the plugin unregisters
- registrations follow the plugin registry + the side-card settings: a
type registered later (external plugin) gets its native type too, a
disabled type is unregistered (absent from the guide, refuses opens)
Adapter (src/client/native/tab-adapter.tsx):
- one synthetic plugin `SidebarTab` per native tab id, minted from the
native record + navigation params (or the descriptor's own createTab,
which a native guide open bypasses), kept live across navigations and
updateTab writes
- the explorer's per-tab expansion/reveal sets and the per-kind instance
counter, which the plugin's own layout used to hold
Service routing (src/client/service.ts):
- `setSurface` installs the native write face; openTab/openFile land
natively (a path becomes a resource address, a path-less editor open
becomes the `files` page, the rest carry the seed as navigation params),
closeTab/updateTab/activateTab recognise native tab ids
- new optional `OpenTabSeed.target` ('right' default, 'bottom' keeps the
bottom workbench); the bottom panel's own + menu and auto-terminal pass
'bottom'
- src/client/native/surface.ts probes the controller's per-session writes
(`openTabIn`/`openResourceIn`/`closeIn`, not in `ISidebarRight`) and
queues opens for a session whose surface is not mounted yet
Right panel retired behaviourally (Phase 3 deletes the code): Sidebar
renders its bottom workbench only, keeps the panel collapsed, and hides
the right-panel toggle. The panel-specific e2e specs are removed; the
mount lane now drives the native guide (tab sweep), the plugin explorer
inside the native pane (editor/mermaid/README chunks) and the sidechat
host routes.
Verified: pnpm typecheck / lint / build / test (126 files, 1349 tests) /
check:consumer-types, plus pnpm test:mount against a real
@deepseek-ai/dsh@0.1.5-alpha.1 host (7/7).
Move the plugin onto the DSH 0.1.5-alpha.1 baseline (peer floor
^0.1.5-alpha.1, CI pinned to the alpha dist-tag) and fix the two host
contract changes that alpha line introduced.
Baseline:
- all @deepseek-ai/dsh-* peer/dev deps -> 0.1.5-alpha.1; optional peer
@deepseek-ai/dsh-client-ui-sidebar-right (declared, unused until the
native-surface phase); devDep @deepseek-ai/dsh-agent-loop (inbox fold)
- dsh.plugin.json engines >=0.1.5-alpha.1; version 0.19.0-alpha.0;
SIDEBAR_SERVICE_VERSION follows
Live assistant stream (assistant/chunk is gone):
- 0.1.5 publishes in-flight model deltas as process-local
agent/assistant-stream frames instead of durable assistant/chunk events
and embeds the settled stream in assistant/message / assistant/attempt.
src/assistant-live.ts folds the frames into a bounded per-session buffer;
sidechat.events serves the current attempt as `live` rows (replaced every
poll, never a delta) and the transcript renders them; the inherited
in-progress snapshot reads message content / attempt stream / live frames.
Cold session reads (persistence.inspect is gone):
- 0.1.5 reads a stored session through an explicit read handle
(open(id,'read') -> handle.read() -> close()). src/session-store.ts owns
that protocol; every cold read (sidechat transcript, cold-resume preset,
session.cwd, changes replay) goes through it.
Dead code and docs:
- drop the remote.session.openWorkspacePath interception: 0.1.5's chat opens
files through ctx.sidebarRight.openResource instead, so the wrapper had no
caller and would hijack the host's "open in app" gesture; the
interceptOpenPath pref, its settings row and its locale keys go with it
- AGENTS.md §3 + external-plugin-guide.md describe the new baseline
- vitest setup bridges Node's localStorage accessor to jsdom's store (the
pre-existing 26 failures on Node >= 22)
- desktop-layout e2e seeds through host RPC and types into the Lexical
composer (0.1.5 replaced the textarea) instead of driving the directory
picker
Verified: pnpm typecheck / lint / build / test (124 files, 1327 tests) /
check:consumer-types, plus pnpm test:mount against a real
@deepseek-ai/dsh@0.1.5-alpha.1 host (18/18).
Escape the .pdf extension in the op-target regex (the unescaped /.pdf$/
matched any path ending in pdf) and fold the copy-pasted toggle button
into PaneToggle; record the sixth deliverable in the design doc.
Records the five features shipped on this PR (md reading mode incl.
mermaid, highlight expansion, secret redaction, html render mode), their
deliberate boundaries (redaction heuristics, saved-file render
semantics), and the 2026-09-06 standards pass.
- redaction useMemo: one redactText pass per field (the outcome carries
both text and hit; was a hit-scan plus a mask pass), and the banner
condition drops the redundant redactionOn check (hit implies on)
- three copy-pasted header toggles -> PaneToggle; the md reading body
-> MdReadingView (absorbing the mermaid branch + codeLabels)
- redaction preference persisted under the repo's dsh-sidebar:v1: prefix
- render toggle limited to .html/.htm (the editor html viewer's ext set;
.xhtml went beyond any claim)
- tsx/jsx share TS_WORDS/JS_WORDS; CSS word list loses its duplicate
entries and its nonexistent true/false constants
- parseReadContent/parseReadLines share the contentLines() envelope core
(content extraction + Showing-lines filter)
Path layer: patterns now match with a right boundary (must not continue
into [a-z0-9]) and list only store-plural forms (credentials/secrets/
passwords, not tokens), so tokenizer.ts / dev.environment.ts /
style.keys.ts / api-tokens.md are no longer masked whole while
.env.local / credentials.json / mytoken.yaml still are.
Content layer: drop the bare key/auth/token/pass/bearer field names —
they label far more ordinary content (keyboard keys, CSS custom
properties, lexical tokens) than secrets; qualified names stay and
Bearer schemes remain covered by BEARER_RE.
Adds the over-masking regression cases; fixes the spec's source import
extension to .ts (repo convention).
.htmlFrame hardcoded #fff -> var(--dsw-alias-bg-base) (mirrors the editor's
.editorHtml); .redactBanner referenced --dsw-alias-state-warning-primary,
which does not exist in the DSW vocabulary (it is state-warn-*), so the
var() always fell through to a hardcoded #b8860b — use the real token with
no literal fallback.
Every .md preview renders through MarkdownDocument (htmlInfo is ungated),
but its markdown segments passed the raw source into MarkdownText, whose
chat-security stance drops non-http(s) image destinations — so local
markdown-syntax images (, reference-style definitions)
always fell back to alt text. Only HTML <img src> tags were rewritten
(postProcessSanitized).
Apply rewriteLocalImageUrls to each markdown run (segment + appended
reference definitions) inside MarkdownDocument's prepared memo, through
the same media trust fence the sanitized HTML leaves already use.
Idempotent for already-absolute media URLs; remote destinations
untouched. Also correct the stale "non-null only for HTML documents"
comment on TextEditor's htmlInfo.
sidechat.start passed `seed` without `meta.isSeeded: true` +
`inheritedEventCount`, so dsh-session treated the whole seed as the
child's OWN events and the child's Inbox constructor replayed the
parent's `agent/inbox/spliced` history — inheriting whatever input sat
UNCLAIMED in the parent at the click moment:
- next-turn: a queued follow-up behind a still-running turn;
- next-step (the reported long-context case, parent queue empty): a
tool-result context / steering spliced between step boundaries.
The first side prompt then claimed [phantom, boundary, question] — the
stale parent message was sent to the model BEFORE the boundary and
logged as a user/message. The fix passes the exact marker pair the
host's own session.fork uses (api-session-controller); ownEvents() now
stops at the end-seed boundary and the inherited inbox replays empty.
Regression tests run the REAL Session.create + Inbox in both
directions (old shape inherits the phantom, marked shape stays empty);
routes tests pin the two create options. Verified: pnpm typecheck /
test (5 pre-existing unrelated UI-spec failures on main unchanged) /
build.
The host-anchor fallback diagnostic strip inlined hardcoded colors
(#f2a1a1 on #1b1b22). Route them through var() chains like the existing
SideChatView.module.css fallback pattern: text/border read
--dsw-alias-state-error-primary (the codebase's canonical error color,
23 existing usages) and the background reads
--dsw-alias-bg-layer-3,var(--dsw-alias-bg-base,...) (elevated surface,
then page base). The old hexes ride as the chain tails, so with no skin
tokens on the page the bar renders byte-identical to before, while any
--dsw-alias-* skin re-themes it (guide §12: no hardcoded colors). The
guide does not cover this strip and tests/theme.spec.ts asserts
effectiveTokenValue semantics only, so neither needs a follow-up.
Catalog entry names (15 tab + 4 viewer) and the dsh-desktop shell preset
desc were raw hardcoded zh strings rendered straight into the add-plugin
modals and the title-bar scheme dropdown. Mirror the existing desc
pattern exactly: PluginEntry.name / ShellPreset.desc widen to the
i18n-friendly string-or-() =>-string union, the catalogs carry lazy
t('pluginXxxName') / t('presetDshDesktopDesc') lookups, and the
add-plugin modal (render, search filter, aria-labels) plus the scheme
SelectMenu resolve them like every other copy. The plain-string arm of
the union stays as the raw-name fallback, and t() itself never yields
empty text (missing keys render the key), so a dict gap can never blank
a name.
20 new dictionary keys added to zh/en (locales.ts), ja (locales-ja.ts),
and all 18 third-language dictionaries — tests/locales.spec.ts enforces
key-set equality against zh for every shipped dict (precedent: 9cc642b),
so the 'leave other locales-* untouched' shortcut would fail the hard
parity gate; real translations are provided instead.
Spec follow-ups (behavior change: names/descs now render through t()):
- plugin-list.spec.ts: name non-empty assertion resolves the union.
- add-plugin-modal.spec.tsx: sentinel/office name assertions resolve
from the catalog (en locale in tests) instead of pinning zh literals.
- shell-presets.spec.ts: desc non-empty assertion resolves the union.
The viewport-corner toggle cluster sits at top:3 riding the right panel's
tab strip while expanded, and recenters to top:14 over DSH's session
header row when collapsed — an 11px snap on every expand/collapse. Ride
the panel slide's own duration/easing (--ds-transition-duration-slow +
--ds-ease-in-out) so both moves stay in lockstep; prefers-reduced-motion
already clears .toggleCluster transitions.
Graduate from the alpha track to the stable channel:
- Baseline: CI mount pin, dsh.plugin.json engines floor, and every
@deepseek-ai/* peer floor / devDependencies pin move to 0.1.2-rc.1
(published on the `next` dist-tag). alpha.5 -> rc.1 is a pure version
bump in the harness (all package.json "version" lines, zero source
delta), so no code adaptation; every host contract was re-verified
unchanged and the real-host mount smoke passes.
- dsh-client-locale resumed publishing; its trailing-pin exception is
gone — peers and devDeps align at 0.1.2-rc.1 (pnpm peers check clean).
- Version: fold the unpublished v0.18.1-alpha.0 / v0.19.0-alpha.0 bumps
into the v0.18.0 stable release; SIDEBAR_SERVICE_VERSION follows
package.json (asserted by tests/service.spec.ts).
- Docs: README/README_EN badges + channel notes + a full "since v0.17.1"
changelog entry (the two folded alpha numbers are annotated), AGENTS.md
§2/§3 baseline wording, external-plugin-guide version line, and the
e2e-host-protocol header's re-verified list gains 0.1.2-rc.1.
On the Windows lane the @deepseek-ai/cordis link produced TS2307 even
though the link was visibly created: under pnpm its target,
node_modules/@deepseek-ai/cordis, is itself a junction into .pnpm, and a
native symlink pointing AT that junction does not traverse for tsc.
Resolve both link targets with node's native realpath first
(GetFinalPathNameByHandle also expands junctions, which bash-level
resolution may leave untouched), and fail loudly with the link layer
listed when a created link does not resolve to a package root.
node-pty's WindowsTerminal queues resize calls that arrive before the
ConPTY control socket's first data flush (_deferNoArgs). If the pty exits
before the flush, the deferred resize throws 'Cannot resize a pty that
has already exited' inside the socket's 'data' handler — uncatchable by
any caller and fatal to the host process. The Windows CI lane caught
exactly this as an unhandled error while all 1233 tests passed.
Park pre-first-output resizes in our own gate (armed at every spawn
site) and replay them via setImmediate once output has flushed the
deferred list — by then node-pty executes resizes synchronously, so the
exited throw lands in tryResizePty's catch instead of the host. POSIX
terminals resize synchronously (no deferral), so the gate is armed on
Windows only and the other platforms keep their exact behavior.
Also route AgentPtyRegistry.resize through tryResizePty: it previously
called pty.resize bare, missing the just-exited throw guard the tool and
attach-socket paths already had.
First ci-windows run exposed four platform assumptions in the fixtures:
- smoke: the scratch repo inherited the runner's system-wide
core.autocrlf=true (Git for Windows default), so every index restore
smudged LF to CRLF and broke the byte-exact assertions — pin
core.autocrlf=false in the fixture (eol policy is not what these
destructive-op tests are about).
- git / git-worktree: Windows TEMP is the 8.3 short form
(C:\Users\RUNNER~1\...) while git reports long paths, and
realpathSync does not expand 8.3 aliases — switch the canonicalizer
to realpathSync.native (libuv resolves via GetFinalPathNameByHandle,
which does); same trick as the existing macOS /var workaround.
- bundle-route: the ETag memo revalidates on stat (mtime+size) and
Windows updates file timestamps lazily, so the equal-size rewrite
still looked unchanged — change the length too.
- market-manifest: GNU tar (Git Bash) reads the drive-letter colon of
absolute Windows paths as a remote-host spec — run it from the pack
dir with the bare tarball name instead.
The pure-logic code carries real win32 branches (shell parsing, WSL path
projection, the PowerShell installer) that mock-injection tests alone
cannot validate. Mirror the ci job on windows-latest (Git Bash as the
step shell) for typecheck / test / build / consumer-type-surface; the
plugin-mount lane (real DSH + Playwright) stays ubuntu-only.
check-consumer-types.sh: force MSYS winsymlinks:nativestrict so Git
Bash's ln -s creates real symlinks instead of deep-copying the repo
(node_modules + full .git history) — inert on Linux/macOS.
Menghuan1918