The injected instruction persists as a durable user message, so an
unscoped 'Do not call any more tools.' stays in every later request's
history. Scope it: '...in this run; further work waits for the user's
next instruction.' A/B probes on deepseek-v4-pro show the scoped wording
is non-inferior in-turn (4/4 zero tool calls, closing quality unchanged)
and next-turn tool use is unaffected under both wordings; the scoped
form states the instruction's actual lifetime.
- fromRequest placeholder: the last two braces of a consecutive } run now
terminate the placeholder, so patterns may end with a brace quantifier
(bot warning; the truncated pattern could even silently mis-match since
an unclosed { is literal in JS regexes)
- document that derived JSONL entries pass through the same resolution
- widen ToolRunContext/deferContext seam docs beyond composite-only usage
(source JSDoc, README pair, core-data-structures type-equiv blocks)
- pin direct-human blocked as uninstructed, completing the
goal-round/direct-human x complete/blocked test quadrant
A goal round reporting complete or blocked used to conclude the turn at
the tool result, so the model never spoke after the call and sessions
ended on a bare update_goal card. The terminal update now defers one
plugin-sourced <goal_complete>/<goal_blocked> instruction onto its
result asking for a grounded closing message without further tool
calls; the turn then ends through the ordinary no-tool-calls stop.
Direct-human mutations stay uninstructed. Wording chosen by A/B
sampling on deepseek-v4-pro; one extra request per goal lifecycle.
New keyless ACP snapshot goal-wrapup drives the shipped app through
create -> round one -> autonomous complete and pins the injection, the
same-turn closing message, and the completed turn end.
A scripted sidecar cannot know values minted at run time, so terminal
goal updates (which must echo the random goal id) were previously
un-scriptable. Placeholders in scripted entries now resolve against the
request corpus at stream time: last match wins, capture group 1 or the
whole match substitutes, and unmatched/invalid/unterminated patterns
fail loud.
Recheck the listing cancellation signal after each child inspection settles:
a per-child read failing with a diagnostic-mapped code during an abort
previously bypassed the inspection's own checkpoints, letting a cancelled
single-child scan return a successful result.
- listChildren() checks its cancellation signal before starting the corpus
trace and immediately after it resolves, so a pre-aborted or trace-time
abort fails CANCELLED instead of returning a successful empty result.
- Corrected the persistence-repair claims: session query resolves persisted
candidates through the non-mutating inspect() read, so listing is
storage-read-only and a torn child tail is surfaced (shorter valid prefix)
rather than repaired; the Agent Note, README, and JSDoc now state that,
and the zh pair is re-recorded.
- listChildren() takes an optional AbortSignal and rechecks it after every
un-signalled session-query await (the cold-resume cooperative-cancellation
boundary); list_agents forwards exec.signal so the registry's drain of
started tool bodies cannot block on a slow or large catalog.
- The list_agents description now presents running/complete as a stored-record
snapshot and defers deliverability to send_message, matching the ownership-
conflict semantics the service tests pin.
Implements the durable-subagent-catalog RFC: SubagentControlService.listChildren()
enumerates a parent's direct continuable children from one sessionQuery trace,
validates each child's sole subagent/descriptor event (now carrying the durable
creation label), and returns one ordered SubagentListEntry[] with per-child
corrupt/unsupported/unavailable diagnostics. The list_agents tool ships as a
separately loadable plugin of dsh-tool-subagent-control requiring sessionQuery
at load; send_message stays usable without it.
Ziya
ZiyaZhang
imccyu
kingwl
Dudu-0223