docs: acceptance run against core main with C15 — 71/0/2 on two databases

[Weegy]

Re-runs docs/ACCEPTANCE-RUN-2026-08-21.md against byte5ai/omadia main after
#815 (epic #470 C15) landed —
on plain origin/main, no patches, no cherry-picks.

Result

Run	Database	PASS	FAIL	BLOCKED
A	fresh	71	0	2
B	pre-seeded (donor rows + 9 tables, plugin ledger dropped)	71	0	2

Same artifact, same script, both exited 0 (the script's exit code is the FAIL count).

Both open items from the previous run are closed

The 33 rows that only passed on the unmerged C12 branch pass on main.
#807 landed and publicPaths.ts no longer carries the two dev-platform
literals. No plugin change was required and none was made — the artifact under
test is the 0.3.1 the repo already carried.

G7 is closed. The C11 handoff used to arrive after core's pre-activate
migration runner had already written all nine ledger rows, so it could only
report alreadySeeded and the skippedNoWitness alarm was unreachable. C15
moves it into the kernel, ahead of the runner:

[tool-runtime] @omadia/dev-platform: ledger handoff — 9 seeded, 0 already seeded,
  0 left for the migration runner (…)
[@omadia/dev-platform] [dev-platform] migrations: 0 applied, 9 already in ledger

Nine adopted, zero applied — against applied 9 migration(s) then
0 seeded, 9 already seeded before.

The alarm now fires

The one output C11 existed to produce, and the one the previous run could not
reach. With the ledger dropped and a single witnessed schema object removed
(the witness for 0030):

… ledger handoff — 8 seeded, 0 already seeded, 1 left for the migration runner
WARN … the donor ledger records 1 file(s) whose witness says the schema object is
  ABSENT; the migration runner will apply them, which is the repair — confirm this
  is the database you think it is: 0030_dev_job_events_truncated_marker.js
… applied 1 migration(s) … (0030_dev_job_events_truncated_marker.js)

Eight adopted, the unwitnessed file named and applied, the index recreated, the
ledger back to 9 and dev_repos untouched.

Under test

Plugin	origin/main @ ec4a2b75, 0.3.1, clean tree
Artifact	omadia-dev-platform-0.3.1.zip, 537,065 bytes
	sha256 618d8b60b95aeceae7258ea0fe362605a358728382ed6706526be980e6ed6124
Core	byte5ai/omadia @ 8900d7b9 (squash of #815)
@omadia/plugin-api	1.6.0
Suite	1,318 tests, 0 fail, Postgres suites live (#572)

Two traps worth knowing

• npm run package does not gate on npm run build. A tsc failure still
  produced a ZIP — 142,081 bytes against 537,065 for a good build, because the UI
  bundle was missing. It still looked like a plugin package. Never publish an
  artifact from a run whose build exit code you did not read.
• Deleting dist/ without *.tsbuildinfo makes an incremental tsc emit
  nothing, and it surfaces one package later as
  Cannot find module '@omadia/dev-platform-plugin-api'.

Docs only — no source, manifest or migration changes.

---

^{Need help on this PR? Tag @codesmith-bot with what you need. Autofix is disabled.}