MBTI-FULL-AUDIT-30: audit complete Chinese MBTI asset estate

[fermatmind]

What changed

• Registered MBTI-FULL-AUDIT-30 and produced a read-only 52-URL zh-CN MBTI inventory with JSON, Markdown, CSV, and live runtime evidence.
• Classified 43 records for content repair and 9 existing records for verification only; preserved the fixed profile, A/T, and hot cross-type route scope.
• Added a deterministic builder and contract coverage; recorded current llms-full degradation and public API latency as an external sidecar issue.

Why

• Establish an authoritative, reproducible batch baseline before group-scoped content, QA, CMS dry-run, approval, import, verification, performance, and indexability work.

Validation

• node scripts/seo/build-mbti-full-audit-30-inventory-runtime-baseline.mjs --allow-network
• focused Vitest contract
• lint . --max-warnings=0
• typecheck
• full contract suite (4 shards, 2057 tests)
• production build with production API and site URL
• YAML/JSON parse, git diff --check, and exact 10-file scope validation

Deferred

• CMS writes/imports, frontend editorial fallback, sitemap/llms runtime changes, deployment, and GSC mutation remain out of scope.
• The llms-full authority/cache regression and observed API latency are recorded for later dedicated runtime work.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: b9de6b352a

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Measure API latency independently

When the HTML page fetch is slower than the API request, this records the combined Promise.all wall time as api_latency_ms, so the audit can falsely classify public API latency and inflate latency_over_10_seconds_count. Capture timing around fetchJson(apiUrl) separately from the canonical HTML fetch before writing the API latency field.

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Reconcile the dependency before adding this train item

This new item depends on MBTI-GSC-25, but the predecessor entry still has next_task: null in the manifest and its ledger still records status: ready_to_merge with merged_at: null; despite the new state entry claiming GitHub shows it merged, train/dependency checks that read the existing manifest/state will see the dependency as unmerged or stop at GSC-25. Reconcile the MBTI-GSC-25 manifest/state and wire its next task when adding this dependent item.

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 285be02126

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Record the final head SHA in the AUDIT-30 ledger

This entry is marked ready_to_merge after later PR updates, but commit_sha still points to the first pushed commit (b9de...) rather than the final reviewed tree/head that includes the PR-open/ready-to-merge ledger updates and the whitespace fix noted below. When closeout or dependency reconciliation uses this field to verify what passed checks or what merged, AUDIT-30 will be pinned to a stale commit; update it when marking the item ready.

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Register the next train item before linking it

This advances the train to MBTI-PROFILE-NT-31, but I only find that ID in this new next_task/routing text and not as an id: in docs/codex/pr-train.yaml or as a state entry. When the train follows AUDIT-30 after merge, the manifest preflight will stop on a missing PR id; either add the authorized manifest/state entry for the next item or leave next_task unset until it exists.

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Match robots directives exactly

If a page returns a robots tag such as index,nofollow, this substring check still passes because nofollow contains follow, so the audit can mark a nofollow/indexability regression as healthy. Split the robots content into directives and require exact index and follow tokens while rejecting exact blocking directives.

Useful? React with 👍 / 👎.