Both sides, one view — Hermes agent, VPS host, guardrails, and scheduled work.
If these disagree, that is what this page exists to catch.
Generated 2026-11-12T07:00:52
At a glance
27/27
guardrails intact
535
skills installed
68
cron jobs
0
failing jobs
No
disk free
Hermes services
Service
State
Detail
gateway-acct2
BACKOFF
Exited too quickly (process log may have details)
gateway-default
RUNNING
pid 1307, uptime 3 days, 13:36:34
hermes-dashboard
RUNNING
pid 1308, uptime 3 days, 13:36:34
prompt-hub
RUNNING
pid 1309, uptime 3 days, 13:36:34
reckoner-api
RUNNING
pid 1310, uptime 3 days, 13:36:34
reckoner-web
RUNNING
pid 1311, uptime 3 days, 13:36:34
reports-static
RUNNING
pid 1312, uptime 3 days, 13:36:34
Skill wiring guardrails
Protected wiring
Result
Integration points protected
27
Intact
27
Lost / needs restore
none
All protected
Every integration point we wired into the ce-* pipelines is present. Upstream syncs auto-restore anything they overwrite.
Scheduled work
Job
Last status
Fail streak
none — all clean
Awaiting retry (not active failures)
Job
Last recorded
Note
Repo Vault Weekly Review (Sunday email)
error
weekly job — not retried since failure (3d ago); next run 2026-11-15T13:00
Read this before acting
These jobs carry an old error from a previous run but have not been retried yet — most are
weekly. They are listed separately so a resolved problem does not read as a live one. Check
the job's next run before treating it as broken.
VPS host services
Container
Status
host unreachable
Cross-environment sync
Check
Result
Container skill categories
117
Host /root/.hermes categories
0
State
in sync / absent
Clean
no host skill tree — clean
Where your changes live
Everything written from Telegram lands in the Hermes Docker volume
/var/lib/docker/volumes/hermes-webui-gsga_hermes-home/_data and
..._hermes-workspace/_data on the VPS disk — verified by writing a marker in the
container and reading it back from the host filesystem. Nothing is container-ephemeral.