SubagentStartRequest.parent, its core-data type-equiv mirror, and the
ACP backend Agent Note still said the out-of-process backend ignores
request.parent — no longer true: with cwd unconfigured it reads exactly
the session header's cwd, and a parent session without one now rejects
at start. State the one-field read at each owning contract, record the
workspace-resolution decision in the Agent Note, and align the README's
accessibility wording with the search-permission guard.
Automatic compaction fires mid-conversation, right after the loop warmed
the provider's KV cache with the last routed request. The default
summarizer then issued a separate request whose prefix shared nothing
with that warm request — a bespoke summarizer system prompt followed by
the older history flattened to one rendered transcript string — so a
differing first token invalidated the entire cached prefix and every
compaction re-processed the whole replayed history twice.
Move the compaction directive from the FRONT (a fresh system prompt) to
the END (a trailing user message), and replay the last routed request's
own system prompt, tools, message prefix, and shadowed-region messages
verbatim via session.requestHeader() + deriveEventMessage. The auxiliary
call is now a genuine prefix-extension of the warm request, so the
provider reuses the cached tokens up to the trailing instruction.
SummarizationInput carries the replayed prefix instead of a flat string;
the now-unused renderTranscript/renderContentBlocks path is removed with
its spec. Cache reuse is best-effort (head compaction guarantees a hit;
a mid-range compaction or a differently-routed summarizer forgoes it),
correctness is not.
Delete @deepseek-ai/dsh-home and make dsh-paths the sole owner of the
single-root harness home ($DSH_HOME || ~/.dsh). Migrate tool-bash,
skill-local, and agent-spine-demo off dsh-home, and fold telemetry's
divergent globalConfigDir onto the shared resolver, dropping its second
XDG/APPDATA policy and the deepseek-harness namespace so the anonymous
id lives under the harness home. Add dshHomeDisplay() for symbolic
user-facing paths, replacing workspace-context's bespoke check.
The master merge brought the tool registry's per-scope restriction layer
(tools.restrict), which makes dsh-mode's prepend assemble filter a duplicate
enforcement shape: it re-implemented the registry's SDK-section rendering
(renderToolsSdk + the RUN_CODE_NAME exclusion) and hid the exit tool from
prompts only — dispatch stayed open and the execute-time re-check was the
real gate. The service now reconciles a per-agent deny restriction on
agent.ctx at agent/created and at every boundary flush, so wire schemas, the
Code Mode tools:sdk section, AND dispatch resolve exit_plan_mode through the
one registry view (a default-mode call answers UNKNOWN_TOOL, byte-identical
to a no-dsh-mode deployment). The execute-time folded-mode re-check stays as
defense in depth for a direct foreign mode/set append no boundary has
reconciled yet.
Two zero-consumer surfaces removed per the pre-release stance:
- AgentOptions.mode creation seeding (declaration merge + agent/created
listener); a caller selects through set() before the first turn, and the
deferred subagent inheritance returns together with its consumer.
- The dropped-definition boundary notice (droppedNoticed + narration):
custom mode definitions have no production consumer, so nothing can be
dropped; fold-to-default degradation is unchanged.
The stdio removal had left plan mode ACP-only while docs still claimed a
/mode command. dsh-mode now registers /mode on the plugin-owned command
registry through an optional ctx.inject(['commands']) child (type-only peer
edge on dsh-commands), so the TUI and the ACP slash-command surface both
gain it; examples/tui-agent composes dsh-mode. ACP/TUI expected outputs
refreshed keyless for the available_commands_update delta.
Docs updated in place (mode READMEs, the plan-mode Agent Note's realization
sections); catalogs and graphs regenerated.
pnpm run doc-sync was a serial &&-chain of 24 pnpm run subcommands; the
24 script bodies finish in ~34s while the chain takes ~3min, all wrapper
overhead. Add a doc-sync mode to scripts/run-gates.ts expanding to
docSyncLeafGates() (concurrency capped at 4 like pre-push: several doc
gates each build a full ts.Program) and point the package script at it,
making the leaf list the single source of truth for the member set.
Consolidating the two lists surfaced drift: verify-cordis-api joined the
package.json chain with the runtime API catalog but was never added to
docSyncLeafGates, so CI and pre-push never gated that catalog's
freshness. Add the missing leaf.
Tianyi Cui
Yichen Jiang
NI0317
Turtle
Chinesezjc
kingwl
ZiyaZhang
Dudu-0223