okf-cli-0.5.0.0: help/trust.md
TRUST, STALENESS, AND PROVENANCE
Two commands report what a bundle says about how far its content can be
trusted. Neither stores anything: every value is derived from frontmatter on
each run, and nothing is read that the bundle does not say.
REPORTING TRUST TIERS
okf trust BUNDLE
One aligned row per concept, ordered by concept ID:
aggregates/invoice human-reviewed stable ok
aggregates/order machine-confirmed stable ok
commands/issue-invoice unverified stable ok
mappings/ordering-to-billing machine-confirmed stable stale since 2026-07-01
policies/reserve-stock unverified draft ok
The four columns are the concept ID, the derived trust tier, the status
field, and staleness. The ID column is padded to the longest ID in the
bundle, so the width shifts when a bundle gains a deeper concept.
TIERS
unverified The concept declares no verified entry.
machine-confirmed Verified, but every verified[].by is a non-human actor.
human-reviewed Some verified[].by uses the human: prefix.
The prefix is the only thing separating the last two. The tier comes from
verified, never from generated: who wrote a concept need not be who confirmed
it.
STATUS AND STALENESS
status shows stable for a concept that declares none, because an absent
status means stable. The other values are draft, deprecated, and superseded.
The staleness column reads "ok" both for a concept with no stale_after and
for one whose deadline has not arrived -- okf does not claim a concept is
fresh, only that nothing says otherwise. A passed deadline prints
"stale since DATE", and a stale_after that is not a YYYY-MM-DD date prints
"unparseable stale_after VALUE" rather than being silently treated as fresh.
Staleness is computed against today, so this command's output changes with
the date even when the bundle does not. It is advisory: okf never refuses a
stale concept.
Staleness here is not the log staleness okf validate and okf log
--check-stale report. That one compares a concept's generated date against
the nearest log.md; this one reads the concept's own stale_after deadline.
LISTING RECORDED PROVENANCE
okf sources BUNDLE
The provenance each concept records, with the credibility signals that frame
it:
aggregates/order
ddd-schema mori://shinzui/mori
author human:nadeem, used 40 times in 2026-01-01..2026-06-18, modified 2026-05-02
ubiquitous-language all order-domain terms agreed in the ordering team's glossary reviews
author human:nadeem, used 6 times in 2026-03-01..2026-06-10, modified 2026-06-10
Concepts with no sources are skipped entirely, so the report shows only what
has provenance. Concepts are ordered by ID, so the output is stable and
diffable in a pipeline.
Entries print in the order the document declares them and are never sorted or
ranked by usage_count. A count is a coarse signal -- read it as liveness and
trend, not as a score -- and a ranked listing would imply a precision it does
not carry.
The window shown after a count is the EFFECTIVE one: an entry's own
usage_window where it has one, the document-scope usage_window otherwise,
which is why two entries on one concept can show different ranges.
A second line is printed only for the signals an entry actually has. An entry
with no id prints "(no id)" in the label column, because an id is optional and
matters only when the body cites the entry with a footnote.
A source's resource is deliberately never path-checked: it names either a
followable artifact or a scope descriptor such as "all queries in project X".
SEE ALSO
okf help format The generated, verified, status, and sources families.
okf help validation Which provenance problems --strict reports.
okf help log Log staleness, which is a different check.