[mldangelo-oai]

Summary

Valid scan usage above 8 KiB could prevent completion or failure from being recorded, leaving a scan marked as running. Remove the serialized-size ceiling so terminal state and complete usage are saved together.

Changes

• Remove the byte-size check while retaining JSON structure, finite-value, token-accounting, ownership, and continuation validation.
• Cover explicit usage on completion and failure, plus usage collected from owned rollout snapshots.

Testing

• All three new regressions failed against the original implementation and passed with the fix.
• Full scan-usage suite: 149 passed, 2 skipped.
• Frozen SDK and MCP dependency installs passed; installed lockfiles match the tracked lockfiles.
• Required plugin checks passed in order: Ruff check, Ruff format check, SDK build:ci, portable source compatibility, and its nine regression tests.
• git diff --check passed.

Risk and rollout

The change accepts larger valid usage records without changing their schema or scan ownership requirements. User-selected cost limits remain enforced. No migration or new configuration is required.

Public disclosure review

• No customer, partner, prospect, or user identities, data, or identifying details are included.
• No credentials, personal data, private source, scan findings, or nonpublic links or tickets are included.
• I reviewed the branch name, title, description, commits, changes, comments, logs, screenshots, attachments, and links for public disclosure.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-23T08:24:16.848571Z	3c3be40	PR opened
🔒 Security Review	✅ Completed 2026-10-23T08:25:09.240719Z	3c3be40	PR opened

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" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[alandelong-oai]

Thanks for fixing large usage persistence. Reviewed both changed files and the completion, failure, usage-reconciliation and cost-limit paths; no actionable issue. Removing the size ceiling preserves the existing usage validation and scan ownership checks.

Fresh checks, builds, types, formatting, 19 focused usage tests and the new failure-state regression passed. Independent base/head controls confirmed large valid records remain intact and malformed records remain rejected. The two completion regressions hit the unchanged ownership guard locally; current hosted macOS CI separately passed all 151 scan-usage cases, including all three new regressions.