HEE-QUOTA(1) HEE Tools HEE-QUOTA(1)
NAME
hee-quota - hee-quota: monitor GitHub API rate-limit pools
(core/graphql/search) and
SYNOPSIS
hee-quota [-h] {status,warn,wait,watch} ...
hee-quota status -- print remaining/limit/reset per pool
hee-quota status --json -- machine-readable
hee-quota warn [--threshold N] -- exit 1 if any pool is below N% remaining
(default 20), otherwise exit 0 silently.
Use as a real pre-flight gate:
hee-quota warn --threshold 15 || exit 1
hee-quota status --pools host -- the HOST as exhaustible pools: memory,
swap, load headroom and agent count, in
the same remaining/limit shape. Costs no
API call and needs no network.
hee-quota watch --pools host -- sample on an interval and STOP ON ITS OWN.
Bounded by design: --samples defaults to
180 and there is no unbounded mode. A
watcher that can run forever is the bug it
was built to catch.
hee-quota wait --pool graphql --min N
-- block (with backoff) until the named
pool has at least N remaining, then
exit 0. Real ETA printed from the
pool's own reset timestamp, not guessed.
positional arguments:
{status,warn,wait,watch}
options:
-h, --help show this help message and exit
DESCRIPTION
give a real, live go/no-go before a bulk operation, instead of finding out
reactively mid-batch.
Real trigger: hit GraphQL exhausted (0/5000, then again at 81/5000) twice
in one session while core REST stayed near-full both times -- confirmed via
`gh api rate_limit` only *after* a batch had already failed partway
through. This is the pre-flight check that should have run first.
Checking rate_limit itself does not consume quota (confirmed live) --
safe to call before every bulk run, not just when something's already
gone wrong.
EXAMPLES
hee quota status
hee quota wait --for 100
Not marked "# ci": it queries the GitHub API rate-limit endpoint, which
needs an authenticated token and is not deterministic on a runner.
HEE-QUOTA(1)