[stackbilt-admin]

Docs-only follow-up to the 1.23.0 release (#123). Two files change: README.md and docs/maintenance/issue-sweep.md. No code or test changes.

README.md

• The pointer to the checkpoint no longer says it lists unreleased fixes. Everything it tracked through 1.23.0 has shipped.
• The Model Catalog section now describes ModelCatalogEntry.explicitOnly, added in 1.23.0. Explicit-only models are served only when a request names them; recommendations and automatic routing skip them. In 1.23.0 that means:
  • the four Claude 5.x models, until Anthropic adapter sends tool_result blocks without tool_use_id, so multi-turn tool use fails #124 fixes Anthropic tool loops;
  • gpt-6-astra, gpt-5.6-sol/-terra/-luna and gpt-5.5, so the OpenAI default stays gpt-4o-mini;
  • SEA-LION v4 on Workers AI.

docs/maintenance/issue-sweep.md

• 1.22.0 is now recorded as released and published to npm, followed by 1.22.1 (fix: correct vendor model IDs and prices against primary docs (v1.22.1) #122) and 1.23.0 (feat: refresh model list against vendor docs, Cloudflare first (#112 phase 2) #123).
• The fix(catalog): verify vendor model IDs and prices and remove fabricated defaults #112 remaining-work note separates the Claude 5.x models (blocked on Anthropic adapter sends tool_result blocks without tool_use_id, so multi-turn tool use fails #124) from the explicit-only OpenAI models (a separate decision, because making them routable would change the gpt-4o-mini default).
• Test counts are measured at 1.23.0 for every listed suite. The missing hook-isolation suite (fix: isolate lifecycle hooks so telemetry cannot replay requests #120) is added, and the unverified historical total is removed.
• Validation evidence is split into what was re-run at 1.23.0 and what dates from the original sweep and was not re-run.
  • Re-run: npm test (789 across 30 files), typecheck, build, test:package and test:worker-import.
  • Not re-run: the sourcemap and whitespace checks.

Checked

• All 20 relative links in the checkpoint and the README's SECURITY.md link resolve.
• The README's explicit-only list matches every catalog entry with explicitOnly: true.
• The version is 1.23.0 everywhere, and the CHANGELOG needed no changes.

Round 1 review (merge after fixes): fixed the wrong #124 reason for the OpenAI models, a false historical test count, stale per-suite counts, the missing hook-isolation suite, and evidence presented as current that was not re-run.

Also posted a release-status update on the roadmap issue #118: #118 (comment)

🤖 Generated with Claude Code