# Public Gooo 0.6.22: exact-source publication and installed observations

Release: https://github.com/kimjooyoon/meta-ontology-go/releases/tag/v0.6.22-dev

Source d3b44fc63a340d825108c97f61eeb18900219b28 was frozen after PR1424's normal
expected-head merge. Its original push-to-dev readiness37923185344, attempt1,
succeeded and the exact original eligibility artifact was verified7/7. Publication
was dispatched once as37923926983 and completed with PUBLISHED/EXACT9/9, nine
public assets and eight checksums. No live original run or watch was restarted.

Original readiness ZIP1060bytes: SHA256
9c05d4912fa70be4aa133ffeafcc4a724fc6e7379efd78a373bf8a8a7b47a1d4.
Original publication ZIP841bytes: SHA256
57fa61c9694767977b0bacc82e870f61a78d646f929cddfeb5cc7fd3a0ff8359.
Raw metadata, digests and verification are retained here.

The public annotated tag, manifest, downloaded metadata/assets, clean binary
identity and installed bytes agree. Public gooo is0.6.22-dev, Go1.27.2,
SDKv0.2.26-experimental with the exact dev source above. Two Korean/English
preflights record explicit zero predictions and candidate tests. Their raw model
input text matches its digest. Eight actual fixed/model construction/replay reports
retain the same selected Go, model/input identity, native4/4 and fields12/12.
Saved runtime replay performs zero fresh inference. Compressed JSON unpacks to
the original bytes; no report was reserialized.

Installed workbench7c37f4b3 used the default PATH gooo with the own-model source
example: target assembly call1, caller4/4, fields12/12, saved replay and Gooo
follow-up advice with unscored0/0 and zero model calls. Fresh affected native race
tests with the public executable and -count=1 passed in18.034s. The previously
cached test invocation is not fresh evidence. These are repeated observations on
the same published four inputs, not independent accuracy samples. Training
exposure is unknown. No new model training or downloads were used.

Main promotion PR1425 and workbench integration PR50 are separately checked.
The source freeze remains active until main also closes; this publication record
does not claim those PRs are merged. Original earlier21 observations remain intact.

## Later closure, 2026-10-09T11:47:37Z

PR50's original full CI succeeded and the expected-head merge produced workbench
main ece4da79. The installed binary from that clean main used default public22
with the existing own model: synthetic partial caller3/4, fields11/12, source
selection12/12, target assembly call1 and unscored follow-up0/0 with calls0.
PR1425's own original canonical CI37923423022 succeeded. Its original v5 proof
and promotion authorization were verified before the normal expected-head merge
to compiler main340982bb6ee7342d70984aa8ea4fb6387d562126. This has the exact
public/dev d3 tree and sole parentf675526de5320136d9e1c35e048bb07c44783244.
Both release-freeze conditions are now complete. Adjacent original main-merged
and workbench-public-merged bundles retain the later evidence.