CodeLedger v1.3.1

[1.3.1] — 2026-05-19

Fixed

• Auto AI review not triggering — _requestAIReview flag was set on the local submission object but never included in the emitSolved() payload; the service worker always saw undefined and skipped the review. Flag is now forwarded correctly.
• "Sync to Ledger" false positive — clicking the button while auto-sync was in progress caused _processSubmission to exit early due to _processingLock, but _manualSync still showed "✓ Synced". The button now waits up to 8 s for the in-progress sync to finish before reporting "✓ Auto-synced", and only shows "✓ Synced" when _processSubmission actually emitted a solve event.
• Floating AI panel position — panel was anchored at bottom: 110px, sitting high above the window bottom. Moved to bottom: 20px to sit at the bottom edge.
• AI system prompt context dropped — buildConversationSystemPrompt had a dead return base statement that prevented the context hints array (problem title, difficulty, request-type behaviour modifiers like "Return useful tests" for /test) from ever reaching the system prompt.
• QoL copy/paste buttons lost on React re-render — LeetCode's React occasionally re-mounts the editor toolbar, silently removing injected buttons. A _maybeReinjectQoL() check is now wired to the existing MutationObserver so buttons are restored within ~600 ms of being removed.

Added

• Monaco language detection in AI panel — _getEditorLanguage() reads the language from monaco.editor.getModels()[0].getLanguageId() and maps it to a human-readable name (e.g. python3 → Python3). Language is now passed to the floating AI context so code blocks include the correct syntax identifier.
• Problem statement auto-injected into AI panel context — _readProblemStatement() reads the problem description from the DOM ([data-track-load="description_content"]) and passes it as problemStatement in the AI chat context, giving the AI full problem context without the user needing /problem.

See CHANGELOG.md for full details.