feat(R04): run every Cowork turn through ConversationApplicationService

R04-T03 — the turn lifecycle, extracted from `core/chat_agent.py::run_cowork`
into `application/conversations/`. The 260-line body mixed the lifecycle (step
budget, cancel checks, guard -> preview -> gate -> execute ordering, sandbox
tidy-up) with the machinery doing each step, and reaching any of it meant
standing up a Qt widget and a worker thread. It is now a plain object driven
through two Protocols and six callables (`turn_runtime.py`), with the concrete
`core/*` wiring confined to `core_runtime_adapter.py` — the same shape R03 used
for routing. Faithful port, not an improvement pass: where the original had a
quirk (the step-ceiling note only merges into the answer when the last message
is the assistant's) the quirk is preserved and commented.

R04-T04 — `ui/cowork_tab.py::build_job` no longer calls run_cowork. It captures
the widget's state at submit time, builds the request via the new
`cowork_turn_request.py` and executes it. `execute(..., messages=...)` hands the
widget's own list over because `_reattach_running_turn` replays from it WHILE
the worker appends and `_finalize_turn` slices it afterwards — a private list
would break both silently.

R04-T05 — `core/task_executors.py`'s cowork branch shares the same engine. All
five unattended-run behaviours stay put (plan reminder, history_ready, History
autosave per assistant message, timeout notice, plan_incomplete_reason), and
`_unattended_prompt` now expresses the load-bearing prefix order in one
readable call instead of three successive rebindings.

Verification: 74 new tests (364 passed, 1 skipped overall; check_imports PASS).
The two that matter most:
- `test_conversation_service_parity.py` runs the same scripted turn through
  run_cowork AND the service and compares the event stream, the resulting
  conversation and the advertised tool list across 7 scenarios;
- `test_task_executor_turn.py` was written BEFORE the migration and passed 8/8
  against the old code, then unchanged against the new.

Known: `ui/cowork_tab.py` (416 -> 455) and `core/task_executors.py` (476 -> 524)
stay above the 400-LOC limit. Both were already over it before this change;
bringing them under needs the R08 / R07 decompositions.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
This commit is contained in:
2026-08-23 13:14:15 +09:00
co-authored by Claude Opus 5
parent 19e6b4deb2
commit 3665135c38
16 changed files with 2452 additions and 48 deletions
+58 -19
View File
@@ -346,19 +346,45 @@ class CoworkTab(ChatPanel):
self._apply_output_folder_label() # picks up edits made via Settings too
def build_job(self, text: str, messages, out_dir):
# Each turn writes into its OWN isolated folder (out_dir) and works on its
# OWN message list, so several turns can run in parallel without clobbering
# each other's files or history. Deliverables are moved up to the session
# Output root when the turn finishes (see _cleanup_turn).
"""This turn's job: a frozen request run through the conversation service.
Since R04-T04 the widget no longer drives the turn loop. Every value a
turn depends on is read HERE, on the UI thread at submit time, and packed
into an immutable ``ConversationExecutionRequest`` — so clicking a
different model or switching workspace mid-answer cannot reach work
already in flight.
"""
output_dir = out_dir or self._session_output_dir()
# The sandbox folder is named by the turn id ('.turns/t3'); with no
# sandbox the session id identifies the turn well enough for the audit log.
turn_id = out_dir.name if out_dir is not None else self.session_id
session_id = self.session_id
title = self.title
project_id = self.project_id
home_output_root = self.workspace_dir()
# Captured at submit time (UI thread): the Admin-defined agent
# preset's instructions, if one is selected in the Agent picker.
agent_prompt = self.admin_agent_prompt()
# Per-workspace Auto-run override wins, else the global "confirm before
# running commands" setting. Frozen now, so a Settings change mid-turn
# cannot flip the rules this turn started under.
confirm_commands = self.ctx.project_confirm_commands()
# What the turn is recorded as running on. A routing override (R03) wins
# over the tab's own picker; '' means the provider's configured default.
# Informational only — an Admin-agent preset builds its own provider
# below, so treat these as the record, not the decision.
provider_id = self._routed_provider or self.ctx.config.active_provider
model = self._routed_model or self._model or ""
def job(worker: AgentWorker):
from ..core.chat_agent import run_cowork
from ..application.conversations.core_runtime_adapter import (
build_cowork_conversation_service,
legacy_event_sink,
)
from ..application.conversations.cowork_turn_request import (
build_cowork_turn_request,
)
from ..application.conversations.turn_runtime import combine_instructions
from ..core.projects import load_project, project_context_text
provider = self.build_provider() # this tab's selected agent/model
@@ -367,23 +393,36 @@ class CoworkTab(ChatPanel):
# built-in MCP server auto-registered while signed in, see
# AppContext._ms365_builtin_connection / mcp_servers/ms365_server.py).
extra_tools, extra_exec = self.ctx.build_mcp_tools()
# Shared project instructions (Claude-Projects style) — refreshed
# each turn so edits in the Workspace screen apply immediately.
proj_ctx = project_context_text(load_project(project_id))
if agent_prompt:
proj_ctx = f"{proj_ctx}\n\n{agent_prompt}" if proj_ctx else agent_prompt
# Shared project instructions (Claude-Projects style) plus the Admin
# agent's persona, refreshed each turn so edits in the Workspace
# screen apply immediately.
instructions = combine_instructions(
project_context_text(load_project(project_id)), agent_prompt)
# Permission Management (Sandbox Security Layer): off by default —
# matches the pre-existing auto-run behavior. Now resolved PER
# WORKSPACE: this project's Auto-run override wins, else the global
# "confirm before running commands" setting (project_confirm_commands).
# matches the pre-existing auto-run behavior. The gate lives on the
# worker because the UI resolves it from the main thread.
gate = None
if self.ctx.project_confirm_commands():
if confirm_commands:
gate = worker.new_gate("confirm", agent_role=agent_roles.COWORK)
run_cowork(provider, messages, output_dir, worker.emit_event,
worker.is_cancelled, title=title,
extra_tools=extra_tools, extra_executor=extra_exec,
project_context=proj_ctx, security_config=self.ctx.config,
gate=gate)
service = build_cowork_conversation_service(
provider, output_dir, worker.emit_event, title=title,
project_context=instructions, extra_tools=extra_tools,
extra_executor=extra_exec, security_config=self.ctx.config,
gate=gate, agent_role=agent_roles.COWORK,
)
request = build_cowork_turn_request(
turn_id=turn_id, session_id=session_id, surface=self.kind,
project_id=project_id, title=title, messages=messages,
provider_id=provider_id, model=model, instructions=instructions,
output_dir=output_dir, home_output_root=home_output_root,
confirm_commands=gate is not None, agent_role=agent_roles.COWORK,
)
# Hand the widget's own list over: _reattach_running_turn replays
# from it while the turn is still running, and _finalize_turn slices
# it afterwards, so the service must append into that very object.
service.execute(request, legacy_event_sink(worker.emit_event),
cancel=worker.is_cancelled, messages=messages)
return {"messages": messages, "turn_dir": str(output_dir)}
return job