Austin Pickett 46370fd2a2 fix(desktop): gate cron-run writes on scheduler ownership, re-evaluated per send ...

The view-only decision for a never-closed cron run keyed on `is_active`,
a 300s activity window. A live run inside a long tool call writes no
heartbeat, reads inactive exactly like a zombie, and was latched
read-only for the rest of the app session.

- Backend: run rows (and the session detail of a cron run) carry
  `scheduler_owned` — the job's in-flight execution is held by a live
  process (pid + start-time fingerprint) and the session started under
  that claim. An older never-closed run of the same job stays a zombie.
- Desktop: the latch becomes a per-run verdict that is re-evaluated on
  every Cron surface poll and, authoritatively, right before any send
  (primary chat and tiles). A run that ticks, is owned, or closes becomes
  writable again; one whose process dies becomes view-only.
- Restored route/tab after a restart: the pre-send refresh recognises a
  cron run by its session id and gates it even though no surface ever
  evaluated it; an unreadable unknown run fails closed.

Refs #88443

2026-10-01 09:44:06 -04:00

..

scripts

…

__init__.py

fix(cron): boot the external worker's PM dependencies before cron.jobs

2026-09-29 03:32:36 +05:30

AGENTS.md

fix(kanban): gate kanban_schedule like the other lifecycle tools

2026-10-01 09:26:00 -04:00

blueprint_catalog.py

…

bot_chat_delivery.py

Merge remote-tracking branch 'origin/main' into ethie/pm-clean

2026-09-20 10:07:50 -04:00

constants.py

refactor(cron): one claim-TTL headroom constant shared by the job store and the ledger

2026-09-20 17:59:57 +05:30

delivery_queue.py

fix(cron): order execution, delivery and incident history by instant, not by text

2026-09-28 19:07:43 +05:30

env_settings.py

…

executions.py

fix(desktop): gate cron-run writes on scheduler ownership, re-evaluated per send

2026-10-01 09:44:06 -04:00

incidents.py

fix(cron): order execution, delivery and incident history by instant, not by text

2026-09-28 19:07:43 +05:30

job_definition.py

fix(cron): normalize a null repeat.completed once in load_jobs

2026-09-26 22:04:17 +05:30

jobs.py

feat(telemetry): gateway platform health, delivery, first-reply latency and cron run metrics

2026-09-28 12:43:03 -07:00

lifecycle_guard.py

fix(cron): remove profile scan ambiguity without dropping long commands

2026-09-30 21:33:08 -04:00

monitor.py

feat(cron): allow Python scripts to use an external interpreter

2026-09-28 02:32:05 +05:30

notepad.py

…

occurrences.py

…

quota_hold.py

fix(cron): fingerprint the unreachable-model retry's schedule expression

2026-09-24 16:59:23 +05:30

scheduler_delivery.py

fix(cron): record the live rejection when the router raises it

2026-09-28 05:28:05 -07:00

scheduler_detached_worker.py

…

scheduler_diagnostics.py

fix: read text files with encoding=utf-8-sig in code merged from main

2026-09-19 01:38:09 -04:00

scheduler_failure_copy.py

fix(agent): an upstream account ban relayed in a 200 stream fails once instead of retrying as an outage

2026-09-23 10:40:08 -07:00

scheduler_ownership.py

fix(cron): release a profile's in-flight claim under the key it registered with

2026-09-21 02:52:15 -07:00

scheduler_preflight.py

fix(cron): satellite route lookup reads the primary's layered config (#121212, salvage #121595)

2026-09-28 05:28:05 -07:00

scheduler_prompt.py

feat(cron): allow Python scripts to use an external interpreter

2026-09-28 02:32:05 +05:30

scheduler_provider.py

fix(cron): detect a routed fire from its home, not a ticker-only marker

2026-09-29 22:57:48 -04:00

scheduler_script.py

fix(cron): overlay the routed no_agent scope inside build_subprocess_env

2026-09-29 22:57:48 -04:00

scheduler_thread.py

fix(cron): the ticker supervisor respawns only a ticker that crashed, not one that returned

2026-09-18 20:07:21 -07:00

scheduler_tick.py

…

scheduler_worker_env.py

fix(cron): stop the external worker when PM refuses its dependency boot

2026-09-29 03:32:36 +05:30

scheduler_worker_failure.py

fix(cron): report acknowledged worker failures without losing diagnostics

2026-09-30 20:57:21 -04:00

scheduler.py

fix(cron): preserve diagnostics when recovery wins a wait timeout

2026-09-30 20:57:21 -04:00

suggestion_catalog.py

…

suggestions.py

…

unreachable_retry.py

refactor(cron): name the retry ladder applicability gate once

2026-09-26 20:52:42 +05:30

worker_bootstrap.py

fix(cron): stop the external worker when PM refuses its dependency boot

2026-09-29 03:32:36 +05:30