Tags

Uptime check from outside AWS every 5 minutes

Nothing outside AWS watched production between deploys. A scheduled
workflow now checks https://hark.fyi/api/health, the legacy
hark.ryan.ceo health endpoint and the homepage, three tries 20 s apart
each; a failed run makes GitHub email whoever last changed the schedule.

Docs limits table fits phones; harkctl 0.8.1 for the updated README

The Free / Legacy Pro limits table had the same 32rem minimum width as
the wide reference tables, so at 390px "10,000" sat off-screen in a
sideways scroll. Plan tables are now `compact` (no minimum width, values
don't wrap); field and route tables keep theirs. Checked at 390px: the
table is 342px wide inside a 342px column, every value visible.

harkctl 0.8.1 carries no code change: npm still shows 0.8.0's README,
which says `npx harkctl auth login`. The README, docs example and the
skill's reviewed version now name 0.8.1.

Verified: all workspace tests pass (harkctl, contracts 42, expo 98,
website 308), typecheck and Biome clean.

Release harkctl: take the pnpm version from packageManager

pnpm/action-setup refuses to run when both its 'version' input and
package.json's packageManager name a version (first release run failed
there). The workflow now relies on packageManager, as production-update
does.