The suite re-imports GoogleAPIService after vi.resetModules() in every test;
under the parallel full run the first import exceeded 5 s and failed
(passes alone in about 1 s).
Add accessible names (and aria-pressed for toggles) to the layout switcher,
overflow menu, selection clear, API key reveal/copy and pricing buttons,
reusing existing i18n keys. Replace the hard-coded English 'Menu' sr-only
text on the account cards with common.openMenu.
responseCache was a plain Map keyed by the JSON of the whole request. Entries
were only removed when the same key was read again after expiring, so every
distinct prompt stayed in memory for the lifetime of the gateway. Replace it
with a TtlCache that prunes expired entries on write and caps the size at 100
entries (TTL unchanged at 60 s).
The cloud ORPC routes (addGoogleAccount, listCloudAccounts, refreshAccountQuota,
syncLocalAccount) returned the full account including access and refresh
tokens, but the renderer never reads them. Blank the two secrets in those
responses; token metadata (expiry, project id) is kept so the schema holds.
Also compare the presented gateway API key with timingSafeEqual.
Delete files with no importers (src/actions/{account,database,window}.ts,
drag-window-region, navigation-menu, ui/navigation-menu, ui/toggle,
ui/toggle-group) and the packages only they or nothing used
(@icons-pack/react-simple-icons, @radix-ui/react-icons,
@sentry/integrations, @radix-ui/react-{navigation-menu,toggle,toggle-group}).
Drop the duplicated Playwright block in .gitignore.
Pin the current behavior with characterization tests (claude-request-mapper,
model-mapping, rate-limit-reason; 91 cases, all green before the refactor),
then split buildContents, buildGenerationConfig, transformClaudeRequestIn,
resolveModelRoute, detectRateLimitReason and computeRetryAfterSeconds into
small helpers. Also declare the emitted toolConfig on GeminiRequest.
- README.md rebuilt from the code: account pool, proxy endpoints, settings paths, verified commands, build/make steps and limitations
- translations get a pointer to the canonical README
- Add LICENSE-MIT for all new code written by evandrodevbr
- Rewrite License section: dual structure (original code CC-BY-NC-SA, new code MIT)
- Add migration notice: project is transitioning to fully open-source MIT
- Rewrite Credits section: accurately attribute what Draculabo created vs what evandrodevbr rewrote
- Remove outdated 'commercial use prohibited' disclaimer
- Applied across all 4 languages (EN, PT-BR, ES, ZH-CN)
- Move Windows SmartScreen workaround instructions to a highly visible warning block at the top of the files
- Add SmartScreen instructions to the Installation / Downloads sections
- Standardize headings, hierarchy, and formatting across all 4 languages (EN, PT-BR, ES, ZH-CN)
- Polish text for better flow and clarity in 'Why Gemini Nexus?' and Tech Stack sections
- Fixes MakerDMG failure (Cannot find module '../build/Release/volume.node')
- Add macos-alias, better-sqlite3, keytar, and electron to trustedDependencies and pnpm.onlyBuiltDependencies
- Also updated pnpm-workspace.yaml allowBuilds directive
- Remove @electron-forge/maker-squirrel dependency
- Squirrel.Windows fails on modern Windows GitHub Actions runners with Win32Exception during zip creation
- We already use MakerWix which produces a much better .msi installer for Windows users
- Add MakerZIP support for win32 to provide a portable alternative alongside the MSI
Vitest workers throw 'Electron failed to install correctly' when backend unit tests indirectly import electron via path/util modules. This mock provides stubs for ipc, app, and BrowserWindow interfaces so happy-dom environments don't crash expecting native binaries.
- Mock service-registry correctly exposing getServiceOptional
- Update tests to expect successful fallback resolutions when services are not yet registered
- Remove obsolete error test that is now correctly handled via ConfigManager fallback
Includes all changes from this release cycle:
- Top header navigation (sidebar removed)
- Accounts screen decomposition and visual redesign
- Dev/prod environment isolation
- Usage Analytics with accurate averages, cost reporting, OpenRouter pricing
- Monitor moved to Usage with per-request token/cost tracking
- Clickable traffic logs with expandable JSON body viewer
- OpenCode IDE config in Quick Setup
- API Proxy reorganization (flattened tabs)
- CI/CD pipeline fixes (release config, Node version, pnpm alignment)
- Repository cleanup (~80MB of binaries removed)
- Delete conflicting .releaserc.json (keep release.config.cjs as single source)
- Fix repo URL from Draculabo/AntigravityManager to evandrodevbr/GeminiNexus
- Fix @semantic-release/git to commit pnpm-lock.yaml (not package-lock.json)
- Normalize Node.js version to 20 across all workflows
- Add packageManager field to package.json for pnpm consistency
- Update AGENTS.md: npm -> pnpm (18 references)
- Update .gitignore: add tools/, thoughts/, binaries, one-shot scripts
- Add opencode to SupportedIde with JSON providers.json output
- Add defaultModel parameter for IDE config generation
- Add model selector dropdown in ToolsTab for IDE configurations
- Use getServiceOptional for graceful degradation when proxy is stopped
- Add fallback IDE config generation using ConfigManager when NestJS is offline
- Update IPC router schema to accept defaultModel parameter
- Add model to logInbound metadata
- Add model, tokensPrompt, tokensCompletion, tokensTotal to logOutbound metadata
- Enrich SSE streaming with token metadata from final usage chunk
- Add metadata parameter to logOutbound method signature
- Data persisted in traffic_logs.metadata_json for per-request cost calculation
- Move MonitorTab and TrafficLogTable from proxy/advanced to usage/
- Add Analytics and Monitor tabs to Usage page
- Remove Monitor tab from API Proxy (flatten to 4 tabs)
- Add Model, Input/Output tokens, and Cost columns to traffic log table
- Make log rows clickable with expandable detail panel (JSON body viewer)
- Fix req/min calculation using actual request time span (MIN/MAX timestamps)
- Add word-wrap for large message bodies in expanded view
- Create CostReportCard with per-model input/output cost breakdown table
- Create ModelPricingDialog for custom pricing configuration
- Add 'Fetch from OpenRouter' button to auto-populate pricing from API
- Add custom_model_pricing to AppConfig (persisted in gui_config.json)
- Refactor calculateCost to accept optional custom pricing overrides
- Support Manual, OpenRouter, and Default pricing sources
- Fix token averaging: calculate over active hours/days only (not full period)
- Enable hourlyData query for all time ranges (not just 24h)
- Rename 'Usage Forecast' to 'Usage Averages' (real data, not projections)
- Add avg/request metric and active hours/days context
- Update terminology: Input/Output Tokens with info tooltips
- Add Estimated Cost stat card with per-model calculation
- Add quota warning alert banner when models exceed 80%
- Add ModelQuotaBar component with color thresholds
- Add IS_DEV_ENVIRONMENT flag to paths.ts for automatic detection
- Dev uses 'Gemini Nexus Dev' app data dir and '.geminiNexus-agent-dev' agent dir
- Dev proxy defaults to port 8046, production stays on 8045
- Move app.setName() before packetLogPath to ensure correct userData path
- Config, DB, logs, and ORPC packets fully isolated between dev and prod
- Rewrite MainLayout from vertical sidebar to horizontal header bar
- Brand left, nav tabs center, status indicator right
- Simplify StatusBar to single inline layout (remove collapsed variant)
- Widen content area from max-w-4xl to max-w-6xl
- Remove sidebar-collapsed localStorage persistence
- Use app.getPreferredSystemLanguages() for reliable OS language detection
instead of relying solely on app.getLocale() which returns Chromium's
internal locale (was returning zh-CN on pt-BR systems)
- Support both Squirrel (%LOCALAPPDATA%) and WiX/MSI (Program Files)
install directories when checking if app runs from expected location
- Extract matchLocaleTag helper for cleaner language resolution logic
- Add getAllExpectedInstallRoots to enumerate all valid install paths
- Normalize Responses API tool format ({name, parameters} at top level)
to Chat Completions format ({function: {name, parameters}})
- Handle 'developer' role as system prompt (OpenAI's newer equivalent)
- Add null/non-array guards to convertOpenAIPartsToAnthropicContent and
extractOpenAITextContent to prevent 'content is not iterable' crashes
- Support input_text, output_text, and reasoning item types in Responses
API input parsing
- Forward reasoning.effort to Claude adaptive thinking config
- Extend JSON schema sanitization blacklist with draft-2020 keywords
(patternProperties, , unevaluatedProperties, etc.)
- Add fallback for untyped items with role/content in Responses API input
- Add diagnostic logging for /v1/responses input conversion
evandro
semantic-release-bot