Review corrections on the observation record:
- `pumpJobOutput` subscribes to `done` once and keeps one pending timer;
racing the same pending promise every poll round retained a reaction
per round for the life of the job.
- The `ctx.jobOutput` effect disposer is async and awaits every open
carrier's `dispose()`, so the fiber reports unloaded only once the old
iterators are quiescent.
- `SessionJob.record: true` replaces `outputTotal`: `append` never commits
a roster change, so the mirrored count was stale while output flowed;
the live offsets ride the observation stream's `opened` anchor.
- The record is documented as a best-effort live preview whose
cross-stream reordering is bounded by the poll interval.
- The web e2e scenarios hold their background job on a test-owned barrier
file instead of a wall-clock sleep, and each is one test rather than a
chain sharing a job id.
Observation (re)opens run on caller stacks — a React expand click, a
carrier retry — whose dynamic context has not declared remote.session, so
the ClientJobOutput wiring resolves the namespace to a concrete value while
the plugin's own context is current, exactly as the activity client did.
Caught by the live-job-stream web e2e, whose goldens (with the
background-job-list pair) now carry the ui-jobs list name; the client-apply
spec gains a jobOutput wiring smoke.
packages/activity/* and packages/api/activity-controller are removed. The
web roster is the session control stream's jobs frames alone; per-job output
observation rides session.observeJob and the client ctx.jobOutput service.
ui-activity returns to its upstream name ui-jobs and renders the single
roster — a row is expandable exactly when its job declared a record
(outputTotal present). Composition rows, dependency-policy exemptions,
tsconfig aliases and references, generated catalogs, and the docs pages
follow; the activity subsystem page is deleted and the jobs page gains the
record contract. The superseding Agent Note records the decision and points
the three prior notes at it.
Candidate rows lead with domain icons instead of localized text
prefixes; pointer and keyboard share one reducer-owned highlight (last
input wins); drillable folder rows reveal a localized Browse-folder +
Tab keycap hint with the library chevron; pending sources render
skeleton bars; the menu spans the composer card. The editable @dir/
text decorates color-only — the domain icon now belongs exclusively to
the settled reference chip. Composer placeholders advertise / and @,
and the zh copy for commands is unified to 指令.
Render reconstructable system prompts at each request-series boundary, preserve series declarations through pre-step wrappers, and keep the presentation and replay snapshots aligned across clients.
Interactive acceptance produced five presentation decisions:
- Header band order is preset → tasks → subagent catalog. The root-session
descendant count moves out of the breadcrumb (header.lineage) into the
actions band (order 30) as SubagentCatalogAction and drops its '/'
separator — a root session has no hierarchy to breadcrumb. Child
sessions keep their root/child switcher unchanged.
- The panel's copy control copies the command (TerminalBlock.copyText),
not the output: a long command ellipsizes in the row, so the control is
its only complete source, and it stays rendered before any output.
- Commands and output lines wrap in full (--dsl-terminal-command-whitespace
/ --dsl-terminal-line-whitespace: pre-wrap), so the output region never
scrolls sideways; the region caps at a fixed height and scrolls
(--dsl-terminal-output-max-height) with maxLines: Infinity disabling the
head/tail fold.
- The panel draws no run-state dot (TerminalBlock.runStateDot: false) —
the row above carries the same state — and reclaims the dot gutter via
--dsl-terminal-gutter, which must target the card element itself because
the variable is declared on it. The panel's left spine is gone with it.
All TerminalBlock changes are opt-ins (props or --dsl-terminal-* variables);
tool cards keep their shipped fold/ellipsis/dot behavior.
Keep recognized code fences syntax-highlighted while assistant text streams. Preserve completed Shiki token lines across chunks, mirror token styles and CRLF handling, and retain plain rendering for unsupported or math-like fences.
Add unit, DOM-parity, and keyless assembled-Web coverage for the streaming-to-settled transition.
Closes#1499
Browser acceptance of the activity panel surfaced the product problem: a
background bash registers in both planes, so the header grew two near-
identical popovers — "background jobs" with lifecycle and duration,
"activities" with live output — listing the same work twice.
The service planes stay orthogonal (jobs: model-visible control with a
consuming read cursor; activities: transient non-consuming observation).
The join happens per row in the UI projection instead:
- dsh-client-ui-activity renders the single merged "tasks" list. Each
jobsBySession row joins the activity carrying its correlation.jobId —
the job supplies identity, lifecycle (stopping exists only there),
duration, and the model-visible detail; the activity supplies the
expandable live-output panel. Standalone activities (workflow runs)
keep their own rows; without the activity registry, job rows still
render and the control degrades to the old job list exactly.
- dsh-client-ui-jobs is deleted (pre-release, no shim); the merged entry
takes its slot order. The job locale namespace folds into activity.
- User-facing copy says "task" (任务): N tasks running / N 个任务进行中.
- Hierarchy: live rows first as two lines (label over kind badge and
status, ticking duration), a Finished divider when both sections
exist, settled rows compacted to one de-emphasized line, and the
output panel hangs behind a spine under its row.
Deliberately not joined: subagent rows stay bare (the subagent panel is
their surface) and foreground commands stay in tool cards.
Both web e2e scenarios now assert one Tasks list; the background-job
scenario's row gained a live-output panel, proving the join end to end.
Agent Note: 2026-08-25-unified-task-list.
Add the activity capability seam: ctx.activities is a pure observation
plane where long-running work (background bash/pwsh, workflow runs) opens a
handle, appends transient output into a byte-offset ring buffer, and settles
exactly once. Nothing model-visible changes: reads are non-consuming and
lossy-by-design (absolute UTF-8 offsets survive eviction and report gaps),
correlation carries {callId, jobId}, and visibility is owner-fenced like jobs.
The seam is strictly optional. Producers tap it through ctx.get('activities')
and contain every observation failure, so an unregistered or broken registry
never affects the underlying job (proven by broken-registry tests).
Ship the full path in one change:
- @deepseek-ai/dsh-activity: Service Definition, pumpActivityOutput pull
substrate, invariant companion; dsh-activity-local: ring-buffer provider
with retention config and owner cleanup.
- shell: ShellProcess.observed exposes non-consuming stdout/stderr readers;
tool-bash/tool-pwsh mirror background jobs into activities.
- tool-workflow: phase/log/agent lines mirrored as a workflow activity.
- api-activity-controller: Typert Remote control (roster baseline+buckets)
and observe (opened/output/status frames, coalesced and byte-capped)
streams plus the client feed service with reference-counted observation.
- ui-activity: session-header action rendering live output through
TerminalBlock, which now shows output while a command is still running.
Docs: activity subsystem page, architecture row, regenerated catalogs, and
the Agent Note (2026-08-24-activity-observation-seam), all bilingual. A
keyless web e2e replays a real run_in_background stream to ARIA goldens.
Regenerate the configuration catalog and module graph after replacing standalone application packages with dsh profile bundles and renaming the private Python carrier. Update built-bin coverage to launch the shipped sdk and acp profiles, assert retired bins stay absent, and keep the Web golden text aligned with the same assembled runtime.
Clarify that headless is a startup profile whose patches freeze after boot. This commit is projection and verification work: it contains no application implementation, and every generated document is produced from source committed earlier in the series.
The turn-error Definition suppressed its node permanently once the owning
turn carried any llm/retry event — a rule from the retired model where a
retry opened a new numbered turn. Retries now run inside the failing turn,
so the suppression hid exactly the exhausted terminal failure it existed
to defer to: spending every transient retry left the conversation with a
neutral collapsed retry row and no error row at all.
Delete the suppression: turn-error matches only turn/start and error-reason
turn/end, and renders whenever its turn recorded a terminal error; the
settled retry chain renders beside it through the separate model-retry
node. The Definition suite now asserts the exhausted-retry error node in
full history, tail-only windows, and after prepending the chain, and a new
keyless live e2e scenario exhausts a scenario-owned two-retry policy and
pins the terminal error row beside the settled retry row (the scaffold
gains replayRetryPolicy so exhaustion runs in milliseconds).
Both stale notes are corrected to the same-turn retry reality, and the new
bug-fix note owns the removal rationale.
Why the record union is one step more abstract than pi-ai's credential
in exactly two places, why a record key names the owning plugin rather
than the provider, why the flow owns the write, and why the interaction
travels with the request instead of a registry. Cross-links the release
fix it supersedes, and states what is still missing: the wire contract
and the Models-page control that would let a human start a login.
The two web e2e goldens regain exactly the openai-codex option line
they lost when the provider was withheld.
A caret Backspace or Delete replaces no selection, so `beforeinput` reports
the bare caret and the previous derivation produced a negative inserted
length and fell back to the ambiguous scan. The delete half of the defect
survived, and the component test missed it by pre-expanding the selection to
the span the engines never report.
The range for a caret delete now comes from the direction `inputType` names
and the number of characters the draft actually lost, measured rather than
assumed to be one, so a grapheme, word, or line deletion sizes correctly.
Only the insert and delete families are recorded; a history replay reports
wherever the caret sits and would name a wrong span while passing every check.
Component tests cover the caret Backspace, Delete, and word-delete gestures,
and an assembled browser scenario drives them as real key presses, which is
the only place an engine's reported range is observable.
Markdown tables were always rendered at natural width, so anything wider
than the 748px message column could only be read through horizontal
scrolling. Following the deepsuite chat TableWrapper treatment: tables
under four columns (and any table inside a blockquote) now fill the column
and wrap cell text down to the cells' minimum readable width, while
four-or-more-column tables keep their natural width behind the wrapper's
horizontal scroll and carry the stable md-table-wide hook. The chat
transcript widens hooked tables past the message column with a pure-CSS
container-query breakout (100cqw against ChatView's scroll container
standing in for deepsuite's JS-measured list width), keeping the table
content aligned with the message column and clamping to neutral when the
transcript is narrower than the column.