personal memory agent
chat: generalize dispatch to talents
+85
-74
+3
-3
apps/chat/tests/test_routes.py
+3
-3
apps/chat/tests/test_routes.py
···
120
120
use_id="use-1",
121
121
text="sol reply",
122
122
notes="full note",
123
-
requested_exec=False,
123
+
requested_target=None,
124
124
requested_task=None,
125
125
)
126
126
append_chat_event(
···
182
182
use_id="use-5",
183
183
text="second",
184
184
notes="",
185
-
requested_exec=False,
185
+
requested_target=None,
186
186
requested_task=None,
187
187
)
188
188
···
213
213
use_id="use-6",
214
214
text="later",
215
215
notes="",
216
-
requested_exec=False,
216
+
requested_target=None,
217
217
requested_task=None,
218
218
)
219
219
+54
-47
convey/chat.py
+54
-47
convey/chat.py
···
33
33
34
34
chat_bp = Blueprint("chat", __name__, url_prefix="/api/chat")
35
35
36
-
MAX_ACTIVE_EXECS = 2
36
+
MAX_ACTIVE_TALENTS = 2
37
37
MAX_LOOP_RETRIES = 3
38
38
DEFAULT_STREAM_LIMIT = 200
39
39
MAX_STREAM_LIMIT = 1000
···
46
46
_current_chat_use_id: str | None = None
47
47
_current_chat_state: dict[str, Any] | None = None
48
48
_queued_trigger: dict[str, Any] | None = None
49
-
_active_execs: dict[str, dict[str, Any]] = {}
49
+
_active_talents: dict[str, dict[str, Any]] = {}
50
50
_last_use_id = 0
51
51
_recovery_day: str | None = None
52
52
_runtime: "ChatRuntimeState | None" = None
···
226
226
raw_chat_use_id = str(_current_chat_state.get("raw_use_id") or "")
227
227
if use_id == raw_chat_use_id:
228
228
logical_use_id = _current_chat_use_id
229
-
elif use_id in _active_execs:
230
-
logical_use_id = str(_active_execs[use_id]["chat_use_id"])
229
+
elif use_id in _active_talents:
230
+
logical_use_id = str(_active_talents[use_id]["chat_use_id"])
231
231
232
232
if logical_use_id is None:
233
233
return
···
279
279
next_info = _clear_current_locked()
280
280
else:
281
281
message_text = parsed["message"] or ""
282
-
requested_exec = parsed["talent_request"] is not None
282
+
requested_target = "exec" if parsed["talent_request"] else None
283
283
requested_task = (
284
284
parsed["talent_request"]["task"]
285
285
if parsed["talent_request"]
···
290
290
use_id=logical_use_id,
291
291
text=message_text,
292
292
notes=parsed["notes"],
293
-
requested_exec=requested_exec,
293
+
requested_target=requested_target,
294
294
requested_task=requested_task,
295
295
)
296
296
_current_chat_state["retry_count"] = 0
297
297
_current_chat_state["raw_use_id"] = None
298
-
if requested_exec:
299
-
active_exec_count = _active_exec_count_for_today_locked()
300
-
if active_exec_count >= MAX_ACTIVE_EXECS:
298
+
if requested_target:
299
+
active_talent_count = _active_talent_count_for_today_locked()
300
+
if active_talent_count >= MAX_ACTIVE_TALENTS:
301
301
_current_chat_state["trigger"] = {
302
302
"type": "synthetic-max-active",
303
303
"reason": MAX_ACTIVE_REASON,
···
305
305
synthetic_use_id = _reserve_use_id_locked()
306
306
_current_chat_state["raw_use_id"] = synthetic_use_id
307
307
next_info = _build_spawn_info_locked(logical_use_id)
308
-
elif _exec_loop_count_locked() >= MAX_LOOP_RETRIES:
308
+
elif _talent_loop_count_locked() >= MAX_LOOP_RETRIES:
309
309
append_chat_event(
310
310
"chat_error",
311
311
reason=CHAT_TROUBLE_REASON,
···
317
317
}
318
318
next_info = _clear_current_locked()
319
319
else:
320
-
exec_use_id = _reserve_use_id_locked()
321
-
_active_execs[exec_use_id] = {
320
+
talent_use_id = _reserve_use_id_locked()
321
+
_active_talents[talent_use_id] = {
322
322
"chat_use_id": logical_use_id,
323
+
"target": requested_target,
323
324
"task": requested_task,
324
325
"location": dict(_current_chat_state["location"]),
325
326
}
326
327
append_chat_event(
327
328
"talent_spawned",
328
-
use_id=exec_use_id,
329
-
name="exec",
329
+
use_id=talent_use_id,
330
+
name=requested_target,
330
331
task=requested_task,
331
-
started_at=int(exec_use_id),
332
+
started_at=int(talent_use_id),
332
333
)
333
334
next_info = {
334
-
"kind": "exec",
335
+
"kind": "talent",
335
336
"logical_use_id": logical_use_id,
336
-
"use_id": exec_use_id,
337
+
"target": requested_target,
338
+
"use_id": talent_use_id,
337
339
"task": requested_task,
338
340
"context": parsed["talent_request"].get("context") or {},
339
341
"location": dict(_current_chat_state["location"]),
···
356
358
}
357
359
next_info = _clear_current_locked()
358
360
359
-
elif use_id in _active_execs:
360
-
exec_state = _active_execs.pop(use_id)
361
-
logical_use_id = str(exec_state["chat_use_id"])
361
+
elif use_id in _active_talents:
362
+
talent_state = _active_talents.pop(use_id)
363
+
logical_use_id = str(talent_state["chat_use_id"])
362
364
summary = str(message.get("result") or "").strip()
363
365
append_chat_event(
364
366
"talent_finished",
365
367
use_id=use_id,
366
-
name="exec",
368
+
name=str(talent_state["target"]),
367
369
summary=summary,
368
370
)
369
371
if (
···
373
375
_current_chat_state["trigger"] = {
374
376
"type": "talent_finished",
375
377
"use_id": use_id,
376
-
"name": "exec",
378
+
"name": str(talent_state["target"]),
377
379
"summary": summary,
378
380
}
379
381
_current_chat_state["raw_use_id"] = _reserve_use_id_locked()
···
407
409
)
408
410
error_payload = {"use_id": logical_use_id, "reason": CHAT_TROUBLE_REASON}
409
411
next_info = _clear_current_locked()
410
-
elif use_id in _active_execs:
411
-
exec_state = _active_execs.pop(use_id)
412
-
logical_use_id = str(exec_state["chat_use_id"])
412
+
elif use_id in _active_talents:
413
+
talent_state = _active_talents.pop(use_id)
414
+
logical_use_id = str(talent_state["chat_use_id"])
413
415
reason = str(message.get("error") or CHAT_TROUBLE_REASON)
414
416
append_chat_event(
415
417
"talent_errored",
416
418
use_id=use_id,
417
-
name="exec",
419
+
name=str(talent_state["target"]),
418
420
reason=reason,
419
421
)
420
422
if (
···
424
426
_current_chat_state["trigger"] = {
425
427
"type": "talent_errored",
426
428
"use_id": use_id,
427
-
"name": "exec",
429
+
"name": str(talent_state["target"]),
428
430
"reason": reason,
429
431
}
430
432
_current_chat_state["raw_use_id"] = _reserve_use_id_locked()
···
443
445
if not _spawn_chat_generate(action):
444
446
_handle_chat_failure(action["logical_use_id"], CHAT_TROUBLE_REASON)
445
447
return
446
-
if action.get("kind") == "exec":
447
-
if not _spawn_exec(action):
448
-
_handle_exec_spawn_failure(action)
448
+
if action.get("kind") == "talent":
449
+
if not _spawn_talent(action):
450
+
_handle_talent_spawn_failure(action)
449
451
450
452
451
453
def _spawn_chat_generate(action: dict[str, Any]) -> bool:
···
477
479
return True
478
480
479
481
480
-
def _spawn_exec(action: dict[str, Any]) -> bool:
482
+
def _spawn_talent(action: dict[str, Any]) -> bool:
481
483
from convey.utils import spawn_agent
482
484
483
-
prompt = _build_exec_prompt(
485
+
prompt = _build_talent_prompt(
486
+
action["target"],
484
487
action["task"],
485
488
action["context"],
486
489
action["location"],
···
493
496
}
494
497
use_id = spawn_agent(
495
498
prompt=prompt,
496
-
name="exec",
499
+
name=action["target"],
497
500
provider=None,
498
501
config=config,
499
502
use_id=action["use_id"],
···
504
507
return True
505
508
506
509
507
-
def _handle_exec_spawn_failure(action: dict[str, Any]) -> None:
510
+
def _handle_talent_spawn_failure(action: dict[str, Any]) -> None:
508
511
next_info: dict[str, Any] | None = None
509
512
with _state_lock:
510
-
_active_execs.pop(str(action["use_id"]), None)
513
+
_active_talents.pop(str(action["use_id"]), None)
511
514
append_chat_event(
512
515
"talent_errored",
513
516
use_id=action["use_id"],
514
-
name="exec",
517
+
name=action["target"],
515
518
reason=CHAT_TROUBLE_REASON,
516
519
)
517
520
if _current_chat_use_id == action["logical_use_id"] and _current_chat_state:
518
521
_current_chat_state["trigger"] = {
519
522
"type": "talent_errored",
520
523
"use_id": action["use_id"],
521
-
"name": "exec",
524
+
"name": action["target"],
522
525
"reason": CHAT_TROUBLE_REASON,
523
526
}
524
527
_current_chat_state["raw_use_id"] = _reserve_use_id_locked()
···
616
619
)
617
620
618
621
619
-
def _active_exec_count_for_today_locked() -> int:
622
+
def _active_talent_count_for_today_locked() -> int:
620
623
return len(reduce_chat_state(_today_day())["active_talents"])
621
624
622
625
623
-
def _exec_loop_count_locked() -> int:
626
+
def _talent_loop_count_locked() -> int:
624
627
events = read_chat_events(_today_day())
625
628
count = 0
626
629
for index in range(len(events) - 1, -1, -1):
···
630
633
break
631
634
if kind != "sol_message":
632
635
continue
633
-
if not event.get("requested_exec"):
636
+
if not event.get("requested_target"):
634
637
continue
635
638
636
639
previous = events[index - 1] if index > 0 else None
···
679
682
}
680
683
681
684
682
-
def _build_exec_prompt(
685
+
def _build_talent_prompt(
686
+
target: str,
683
687
task: str,
684
688
context_hints: dict[str, Any],
685
689
location: dict[str, str],
···
703
707
history_lines.append(f"**Sol**: {event['text']}")
704
708
if history_lines:
705
709
parts.append("Recent chat:\n" + "\n".join(history_lines[-6:]))
710
+
711
+
if target != "exec":
712
+
parts.append(f"Target: {target}")
706
713
707
714
return "\n\n".join(parts)
708
715
···
784
791
return None
785
792
786
793
latest_sol: dict[str, Any] | None = None
787
-
exec_state: dict[str, Any] | None = None
794
+
talent_state: dict[str, Any] | None = None
788
795
chat_error: dict[str, Any] | None = None
789
796
spawned_task: str | None = None
790
797
···
796
803
chat_error = event
797
804
elif kind == "talent_spawned" and str(event.get("use_id")) == use_id:
798
805
spawned_task = event.get("task")
799
-
exec_state = {"state": "active", "task": spawned_task}
806
+
talent_state = {"state": "active", "task": spawned_task}
800
807
elif kind == "talent_finished" and str(event.get("use_id")) == use_id:
801
-
exec_state = {
808
+
talent_state = {
802
809
"state": "finished",
803
810
"summary": event.get("summary", ""),
804
811
"task": spawned_task,
805
812
}
806
813
elif kind == "talent_errored" and str(event.get("use_id")) == use_id:
807
-
exec_state = {
814
+
talent_state = {
808
815
"state": "errored",
809
816
"reason": event.get("reason", ""),
810
817
"task": spawned_task,
···
813
820
with _state_lock:
814
821
if _current_chat_use_id == use_id:
815
822
task = None
816
-
if latest_sol and latest_sol.get("requested_exec"):
823
+
if latest_sol and latest_sol.get("requested_target"):
817
824
task = latest_sol.get("requested_task")
818
825
return {"state": "active", "task": task}
819
826
···
827
834
"state": "finished",
828
835
"summary": latest_sol.get("text", ""),
829
836
}
830
-
return exec_state
837
+
return talent_state
831
838
832
839
833
840
def _read_talent_log(use_id: str) -> dict[str, Any] | None:
+2
-2
convey/chat_stream.py
+2
-2
convey/chat_stream.py
···
28
28
"use_id",
29
29
"text",
30
30
"notes",
31
-
"requested_exec",
31
+
"requested_target",
32
32
"requested_task",
33
33
),
34
34
"talent_spawned": ("use_id", "name", "task", "started_at"),
···
122
122
"use_id": event["use_id"],
123
123
"text": event["text"],
124
124
"notes": event["notes"],
125
-
"requested_exec": event["requested_exec"],
125
+
"requested_target": event["requested_target"],
126
126
"requested_task": event["requested_task"],
127
127
}
128
128
continue
+1
-1
talent/chat.md
+1
-1
talent/chat.md
···
49
49
- If the owner is asking about imports, naming, or system readiness, answer plainly from the supplied context.
50
50
- Request exec only when answering well requires deeper lookup, synthesis, or tool use.
51
51
52
-
## When To Dispatch Exec
52
+
## When To Dispatch Talents
53
53
54
54
Set `talent_request` only when the owner needs work that cannot be answered well from the supplied digest, chat history, active routines, and trigger context alone.
55
55
+1
-1
talent/chat_context.py
+1
-1
talent/chat_context.py
···
435
435
if not active_talents:
436
436
return ""
437
437
438
-
lines = ["## Active Execs\n"]
438
+
lines = ["## Active Talents\n"]
439
439
for talent in active_talents:
440
440
started_at = _format_started_at(talent.get("started_at"))
441
441
line = f"- **{talent.get('name', 'exec')}** — {talent.get('task', '')}"
+1
-1
tests/baselines/api/sol/preview.json
+1
-1
tests/baselines/api/sol/preview.json
···
1
1
{
2
-
"full_prompt": "## Instructions\n\n## Available Facets\n\n- **Capulet Industries** (`capulet`)\n Capulet Industries enterprise division\n - **Capulet Industries Entities**: Tybalt Capulet; Juliet Capulet; Paris Duke; Nurse Angela; Capulet Industries\n - **Capulet Industries Activities**:\n - Meetings\n - Coding\n - Browsing\n - Email\n - Messaging\n - AI Conversation\n - Writing\n - Reading\n - Video\n - Gaming\n - Social Media\n - Planning\n - Productivity\n - Terminal\n - Design\n - _and 1 more activities_\n\n- **Empty Entities Test** (`empty-entities`)\n - **Empty Entities Test Activities**:\n - Meetings\n - Coding\n - Browsing\n - Email\n - Messaging\n - AI Conversation\n - Writing\n - Reading\n - Video\n - Gaming\n - Social Media\n - Planning\n - Productivity\n - Terminal\n - Design\n - _and 1 more activities_\n\n- **Full Featured Facet** (`full-featured`)\n A facet for testing all features\n - **Full Featured Facet Entities**: First test entity; Second test entity; Third test entity with description\n - **Full Featured Facet Activities**: Meetings; Coding; Custom Activity; Email; Messaging\n\n- **Minimal Facet** (`minimal-facet`)\n - **Minimal Facet Activities**:\n - Meetings\n - Coding\n - Browsing\n - Email\n - Messaging\n - AI Conversation\n - Writing\n - Reading\n - Video\n - Gaming\n - Social Media\n - Planning\n - Productivity\n - Terminal\n - Design\n - _and 1 more activities_\n\n- **Montague Tech** (`montague`)\n Montague Tech startup operations\n - **Tester's Role**: CTO and co-founder of Montague Tech. Visionary full-stack engineer.\n - **Montague Tech Entities**: Mercutio Escalus; Benvolio Montague; Juliet Capulet; Verona Platform; Mesh Routing; Montague Tech; Prince Escalus; Verona Ventures; Rosaline Prince; Balcony App; Schema Bridge; Friar Lawrence; Balthasar Davi\n - **Montague Tech Activities**: Engineering; Meetings; Email; Messaging\n\n- **Priority Test** (`priority-test`)\n - **Priority Test Activities**:\n - Meetings\n - Coding\n - Browsing\n - Email\n - Messaging\n - AI Conversation\n - Writing\n - Reading\n - Video\n - Gaming\n - Social Media\n - Planning\n - Productivity\n - Terminal\n - Design\n - _and 1 more activities_\n\n- **Test Facet** (`test-facet`)\n A test facet for validating functionality\n - **Test Facet Entities**: John Smith; Acme Corp; API Optimization; Bob Wilson; Dashboard Redesign; Docker; Jane Doe; PostgreSQL; Tech Solutions Inc; Visual Studio Code\n - **Test Facet Activities**:\n - Meetings\n - Coding\n - Browsing\n - Email\n - Messaging\n - AI Conversation\n - Writing\n - Reading\n - Video\n - Gaming\n - Social Media\n - Planning\n - Productivity\n - Terminal\n - Design\n - _and 1 more activities_\n\n- **Verona** (`verona`)\n Cross-company Verona Platform collaboration\n - **Tester's Role**: Co-lead of the Verona Platform joint venture from Montague Tech.\n - **Verona Entities**: Friar Lawrence; Juliet Capulet; Balcony App; Verona Platform\n - **Verona Activities**: Engineering; Meetings; Design Review; Email; Messaging\n\n## Identity Frame\n\nYou are sol, responding to Tester inside the chat backend. You are not the research worker and you do not have tools in this step. Work only from the context already provided to you.\n\n## Current Digest\n\n$digest_contents\n\n$location\n\n$trigger_context\n\n$active_talents\n\n$active_routines\n\n$routine_suggestion\n\n## Tonal Range\n\nMatch the owner's tone and stakes:\n- Be direct and brief for simple replies.\n- Be warm when the owner is sharing something difficult or personal.\n- Be analytical when the owner needs synthesis or a plan.\n- Be challenging only when there is a clear pattern worth naming.\n\n## Routine Etiquette\n\n- If a routine suggestion appears in context, mention it once and only at the end.\n- Do not raise routine suggestions on machine-driven follow-ups unless the context explicitly includes one.\n- Do not mention internal systems, hooks, or prompt assembly.\n\n## Import And Naming Awareness\n\n- If the owner is asking about imports, naming, or system readiness, answer plainly from the supplied context.\n- Request exec only when answering well requires deeper lookup, synthesis, or tool use.\n\n## When To Dispatch Exec\n\nSet `talent_request` only when the owner needs work that cannot be answered well from the supplied digest, chat history, active routines, and trigger context alone.\n\nDispatch exec for:\n- Journal exploration across days, entities, or transcripts\n- Multi-step synthesis or research\n- Meeting prep that needs fresh participant or activity lookup\n- Any request that clearly needs tool use or external state inspection\n\nDo not dispatch exec for:\n- Simple acknowledgements\n- Straightforward follow-up chat\n- Routine suggestions already supported by the supplied context\n- Brief guidance that can be answered from the current digest and chat tail\n\n## JSON Contract\n\nReturn exactly one JSON object matching `chat.schema.json`.\n\n- `message`: The owner-facing reply. Use `null` only when you genuinely have no safe or useful message to send.\n- `notes`: Brief internal summary of why you responded this way. Keep it factual and concise. Do not dump long reasoning.\n- `talent_request`: `null` unless exec should be dispatched. When dispatching, include:\n - `task`: the exact work exec should perform\n - `context`: optional structured hints that will help exec start fast\n\n## Output Rules\n\n- Return JSON only.\n- `message` should stand on its own without referring to hidden machinery.\n- If `talent_request` is present, the `message` should still be useful to the owner right now.\n- Prefer no dispatch over a weak or redundant dispatch.",
2
+
"full_prompt": "## Instructions\n\n## Available Facets\n\n- **Capulet Industries** (`capulet`)\n Capulet Industries enterprise division\n - **Capulet Industries Entities**: Tybalt Capulet; Juliet Capulet; Paris Duke; Nurse Angela; Capulet Industries\n - **Capulet Industries Activities**:\n - Meetings\n - Coding\n - Browsing\n - Email\n - Messaging\n - AI Conversation\n - Writing\n - Reading\n - Video\n - Gaming\n - Social Media\n - Planning\n - Productivity\n - Terminal\n - Design\n - _and 1 more activities_\n\n- **Empty Entities Test** (`empty-entities`)\n - **Empty Entities Test Activities**:\n - Meetings\n - Coding\n - Browsing\n - Email\n - Messaging\n - AI Conversation\n - Writing\n - Reading\n - Video\n - Gaming\n - Social Media\n - Planning\n - Productivity\n - Terminal\n - Design\n - _and 1 more activities_\n\n- **Full Featured Facet** (`full-featured`)\n A facet for testing all features\n - **Full Featured Facet Entities**: First test entity; Second test entity; Third test entity with description\n - **Full Featured Facet Activities**: Meetings; Coding; Custom Activity; Email; Messaging\n\n- **Minimal Facet** (`minimal-facet`)\n - **Minimal Facet Activities**:\n - Meetings\n - Coding\n - Browsing\n - Email\n - Messaging\n - AI Conversation\n - Writing\n - Reading\n - Video\n - Gaming\n - Social Media\n - Planning\n - Productivity\n - Terminal\n - Design\n - _and 1 more activities_\n\n- **Montague Tech** (`montague`)\n Montague Tech startup operations\n - **Tester's Role**: CTO and co-founder of Montague Tech. Visionary full-stack engineer.\n - **Montague Tech Entities**: Mercutio Escalus; Benvolio Montague; Juliet Capulet; Verona Platform; Mesh Routing; Montague Tech; Prince Escalus; Verona Ventures; Rosaline Prince; Balcony App; Schema Bridge; Friar Lawrence; Balthasar Davi\n - **Montague Tech Activities**: Engineering; Meetings; Email; Messaging\n\n- **Priority Test** (`priority-test`)\n - **Priority Test Activities**:\n - Meetings\n - Coding\n - Browsing\n - Email\n - Messaging\n - AI Conversation\n - Writing\n - Reading\n - Video\n - Gaming\n - Social Media\n - Planning\n - Productivity\n - Terminal\n - Design\n - _and 1 more activities_\n\n- **Test Facet** (`test-facet`)\n A test facet for validating functionality\n - **Test Facet Entities**: John Smith; Acme Corp; API Optimization; Bob Wilson; Dashboard Redesign; Docker; Jane Doe; PostgreSQL; Tech Solutions Inc; Visual Studio Code\n - **Test Facet Activities**:\n - Meetings\n - Coding\n - Browsing\n - Email\n - Messaging\n - AI Conversation\n - Writing\n - Reading\n - Video\n - Gaming\n - Social Media\n - Planning\n - Productivity\n - Terminal\n - Design\n - _and 1 more activities_\n\n- **Verona** (`verona`)\n Cross-company Verona Platform collaboration\n - **Tester's Role**: Co-lead of the Verona Platform joint venture from Montague Tech.\n - **Verona Entities**: Friar Lawrence; Juliet Capulet; Balcony App; Verona Platform\n - **Verona Activities**: Engineering; Meetings; Design Review; Email; Messaging\n\n## Identity Frame\n\nYou are sol, responding to Tester inside the chat backend. You are not the research worker and you do not have tools in this step. Work only from the context already provided to you.\n\n## Current Digest\n\n$digest_contents\n\n$location\n\n$trigger_context\n\n$active_talents\n\n$active_routines\n\n$routine_suggestion\n\n## Tonal Range\n\nMatch the owner's tone and stakes:\n- Be direct and brief for simple replies.\n- Be warm when the owner is sharing something difficult or personal.\n- Be analytical when the owner needs synthesis or a plan.\n- Be challenging only when there is a clear pattern worth naming.\n\n## Routine Etiquette\n\n- If a routine suggestion appears in context, mention it once and only at the end.\n- Do not raise routine suggestions on machine-driven follow-ups unless the context explicitly includes one.\n- Do not mention internal systems, hooks, or prompt assembly.\n\n## Import And Naming Awareness\n\n- If the owner is asking about imports, naming, or system readiness, answer plainly from the supplied context.\n- Request exec only when answering well requires deeper lookup, synthesis, or tool use.\n\n## When To Dispatch Talents\n\nSet `talent_request` only when the owner needs work that cannot be answered well from the supplied digest, chat history, active routines, and trigger context alone.\n\nDispatch exec for:\n- Journal exploration across days, entities, or transcripts\n- Multi-step synthesis or research\n- Meeting prep that needs fresh participant or activity lookup\n- Any request that clearly needs tool use or external state inspection\n\nDo not dispatch exec for:\n- Simple acknowledgements\n- Straightforward follow-up chat\n- Routine suggestions already supported by the supplied context\n- Brief guidance that can be answered from the current digest and chat tail\n\n## JSON Contract\n\nReturn exactly one JSON object matching `chat.schema.json`.\n\n- `message`: The owner-facing reply. Use `null` only when you genuinely have no safe or useful message to send.\n- `notes`: Brief internal summary of why you responded this way. Keep it factual and concise. Do not dump long reasoning.\n- `talent_request`: `null` unless exec should be dispatched. When dispatching, include:\n - `task`: the exact work exec should perform\n - `context`: optional structured hints that will help exec start fast\n\n## Output Rules\n\n- Return JSON only.\n- `message` should stand on its own without referring to hidden machinery.\n- If `talent_request` is present, the `message` should still be useful to the owner right now.\n- Prefer no dispatch over a weak or redundant dispatch.",
3
3
"multi_facet": false,
4
4
"name": "chat",
5
5
"title": "Chat"
+3
-3
tests/test_chat_context.py
+3
-3
tests/test_chat_context.py
···
84
84
use_id="use-chat-1",
85
85
text="I can help with that.",
86
86
notes="Responded directly.",
87
-
requested_exec=False,
87
+
requested_target=None,
88
88
requested_task=None,
89
89
)
90
90
append_chat_event(
···
150
150
{"role": "assistant", "content": "I can help with that."},
151
151
]
152
152
assert all("exec spawned" not in msg["content"] for msg in result["messages"])
153
-
assert "## Active Execs" in template_vars["active_talents"]
153
+
assert "## Active Talents" in template_vars["active_talents"]
154
154
assert "Prepare the meeting brief" in template_vars["active_talents"]
155
155
assert "## Trigger Context" in template_vars["trigger_context"]
156
156
assert "Type: owner_message" in template_vars["trigger_context"]
···
229
229
use_id="use-chat-2",
230
230
text="Looking into it.",
231
231
notes="Acknowledged request.",
232
-
requested_exec=False,
232
+
requested_target=None,
233
233
requested_task=None,
234
234
)
235
235
append_chat_event(
+8
-6
tests/test_chat_runtime.py
+8
-6
tests/test_chat_runtime.py
···
14
14
chat_module._current_chat_use_id = None
15
15
chat_module._current_chat_state = None
16
16
chat_module._queued_trigger = None
17
-
chat_module._active_execs.clear()
17
+
chat_module._active_talents.clear()
18
18
chat_module._recovery_day = None
19
19
chat_module._last_use_id = 0
20
20
···
26
26
return journal
27
27
28
28
29
-
def test_chat_result_with_two_active_execs_retriggers_with_max_active_reason(
29
+
def test_chat_result_with_two_active_talents_retriggers_with_max_active_reason(
30
30
tmp_path, monkeypatch
31
31
):
32
32
import convey.chat as chat
···
85
85
sol_messages = [
86
86
e for e in read_chat_events(chat._today_day()) if e["kind"] == "sol_message"
87
87
]
88
-
assert sol_messages[-1]["requested_exec"] is True
88
+
assert sol_messages[-1]["requested_target"] == "exec"
89
89
assert sol_messages[-1]["requested_task"] == "research it"
90
90
91
91
···
117
117
use_id="1713621999999",
118
118
text=f"follow up {index}",
119
119
notes="retrying",
120
-
requested_exec=True,
120
+
requested_target="exec",
121
121
requested_task=f"task {index}",
122
122
)
123
123
···
182
182
"location": {"app": "sol", "path": "/app/sol", "facet": "work"},
183
183
"retry_count": 0,
184
184
}
185
-
chat._active_execs["1713623000001"] = {
185
+
chat._active_talents["1713623000001"] = {
186
186
"chat_use_id": "1713623000000",
187
+
"target": "exec",
187
188
"task": "summarize",
188
189
"location": {"app": "sol", "path": "/app/sol", "facet": "work"},
189
190
}
···
205
206
"location": {"app": "sol", "path": "/app/sol", "facet": "work"},
206
207
"retry_count": 0,
207
208
}
208
-
chat._active_execs["1713624000001"] = {
209
+
chat._active_talents["1713624000001"] = {
209
210
"chat_use_id": "1713624000000",
211
+
"target": "exec",
210
212
"task": "summarize",
211
213
"location": {"app": "sol", "path": "/app/sol", "facet": "work"},
212
214
}
+5
-5
tests/test_chat_stream.py
+5
-5
tests/test_chat_stream.py
···
99
99
use_id="1713626000000",
100
100
text="hello",
101
101
notes="ready",
102
-
requested_exec=False,
102
+
requested_target=None,
103
103
requested_task=None,
104
104
)
105
105
···
296
296
use_id="chat-1",
297
297
text="dispatching",
298
298
notes="planning",
299
-
requested_exec=True,
299
+
requested_target="exec",
300
300
requested_task="compare drafts",
301
301
)
302
302
append_chat_event(
···
343
343
"use_id": "chat-1",
344
344
"text": "dispatching",
345
345
"notes": "planning",
346
-
"requested_exec": True,
346
+
"requested_target": "exec",
347
347
"requested_task": "compare drafts",
348
348
}
349
349
assert reduced["active_talents"] == [
···
402
402
use_id="chat-1",
403
403
text="working",
404
404
notes="",
405
-
requested_exec=False,
405
+
requested_target=None,
406
406
requested_task=None,
407
407
)
408
408
append_chat_event(
···
436
436
use_id="chat-1",
437
437
text="thanks",
438
438
notes="",
439
-
requested_exec=False,
439
+
requested_target=None,
440
440
requested_task=None,
441
441
)
442
442
+6
-4
tests/test_convey_chat.py
+6
-4
tests/test_convey_chat.py
···
26
26
chat_module._current_chat_use_id = None
27
27
chat_module._current_chat_state = None
28
28
chat_module._queued_trigger = None
29
-
chat_module._active_execs.clear()
29
+
chat_module._active_talents.clear()
30
30
chat_module._recovery_day = None
31
31
chat_module._last_use_id = 0
32
32
···
86
86
assert starts and starts[-1]["logical_use_id"] == payload["use_id"]
87
87
88
88
89
-
def test_session_endpoint_reduces_from_chat_stream(chat_client):
89
+
def test_session_endpoint_reduces_from_chat_stream(chat_client, monkeypatch):
90
90
day = "20260420"
91
+
monkeypatch.setattr("convey.chat._today_day", lambda: day)
91
92
append_chat_event(
92
93
"sol_message",
93
94
ts=_ms(2026, 4, 20, 12, 0, 0),
94
95
use_id="1713626000000",
95
96
text="hello",
96
97
notes="ready",
97
-
requested_exec=False,
98
+
requested_target=None,
98
99
requested_task=None,
99
100
)
100
101
append_chat_event(
···
136
137
use_id = str(_ms(2026, 4, 20, 12, 0, 0))
137
138
append_chat_event(
138
139
"sol_message",
140
+
ts=_ms(2026, 4, 20, 12, 0, 0),
139
141
use_id=use_id,
140
142
text="stream reply",
141
143
notes="done",
142
-
requested_exec=False,
144
+
requested_target=None,
143
145
requested_task=None,
144
146
)
145
147