teknium1 27144dd59a fix: the wedged-cron drain check parses jobs.json once per run (review follow-up) ...

get_wedged_job_ids called cron.jobs.get_job (a full jobs.json load) per in-flight
job, and the restart drain invokes it on the event loop every 0.1 s for the whole
wait: 3 running jobs x 600 s = ~18,000 parses. Resolve each run's allowance once
(cached per job_id while it is registered, dropped on release) from a single
load_jobs() call.

2026-09-20 12:24:10 -07:00

..

scripts

…

__init__.py

…

AGENTS.md

fix(kanban): dispatcher blocks a card on the first terminal provider error

2026-09-18 20:34:16 -07:00

blueprint_catalog.py

refactor(skills): live-dashboard becomes an optional skill reconfigured for the Desktop app

2026-09-15 03:35:33 -07:00

bot_chat_delivery.py

fix(cron): degraded-delivery marker is recognised by its record, not its text

2026-09-20 17:55:54 +05:30

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

refactor(cron): keep the suppressed disposition, drop the delivery-manifest ledger rework

2026-09-18 01:43:35 +05:30

env_settings.py

…

executions.py

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

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

incidents.py

fix(cron): remind once per cooldown after the alert-once gate; migrate alerted_at

2026-09-18 09:40:00 -07:00

jobs.py

feat(cron): jobs follow the main agent model at fire time; pinned locks it on request

2026-09-20 09:20:51 -07:00

lifecycle_guard.py

fix(cron): -f patterns must match the gateway cmdline, not any "hermes" substring; ledger names agent-issued kills

2026-09-18 12:49:13 -07:00

monitor.py

…

notepad.py

fix(cron): sqlite_util imports are late so a running scheduler survives an on-disk upgrade

2026-09-13 05:08:29 -07:00

occurrences.py

refactor(cron): probe name covers both bounds; delattr fails loudly; one-line facade import

2026-09-18 12:54:24 +05:30

quota_hold.py

fix(cron): hold fires through a provider's closed usage window instead of re-firing every tick

2026-09-19 10:12:47 -07:00

scheduler_delivery.py

fix(bot-relay): a relayed turn is booked from its turn report at the cap, not killed with its handoff

2026-09-20 11:55:07 -07:00

scheduler_detached_worker.py

fix(cron): terminal-worker reaper lives in scheduler_detached_worker, not the scheduler facade

2026-09-18 09:16:57 -07:00

scheduler_diagnostics.py

fix(cron): a restart-safe worker that dies before its ack reports its own stderr

2026-09-16 17:52:40 -07:00

scheduler_failure_copy.py

fix: WAF-block follow-ups — honest hint, cron action, anthropic debug dump (#24293)

2026-09-19 09:57:21 -07:00

scheduler_preflight.py

fix(cron): a missing-credential preflight verdict names the profile and HERMES_HOME it read

2026-09-19 12:14:32 -07:00

scheduler_prompt.py

fix(cron): record agent-declared cron failures

2026-09-16 17:11:51 -07:00

scheduler_provider.py

fix(cron): sleep to deadline in ticker loops to eliminate clock drift (#114467)

2026-09-18 09:17:29 -07:00

scheduler_script.py

fix(cron): decode cron child output lossily (script + bot-chat delivery)

2026-09-20 00:06:46 -07:00

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

fix(serve): hold work admission through cooperative retirement

2026-09-17 00:34:50 -05:00

scheduler_worker_env.py

fix(cron): pin the worker's PYTHONPATH from the sanitized env in a sibling helper; skip under a wheel install

2026-09-17 08:54:39 -07:00

scheduler.py

fix: the wedged-cron drain check parses jobs.json once per run (review follow-up)

2026-09-20 12:24:10 -07:00

suggestion_catalog.py

…

suggestions.py

refactor(persistence): 24 hand-rolled atomic JSON/text writers go through utils.atomic_json_write / atomic_write_text

2026-09-13 05:07:11 -07:00

unreachable_retry.py

feat(cron): automatic bounded re-runs when a fire never reached the model

2026-09-12 22:13:33 -07:00