Duplicate release-assets/OmniRoute.exe glob caused softprops/action-gh-release
to attempt a second upload of the same asset, triggering a 404 Not Found error
on the GitHub release asset update API. The file is already covered by the
*.exe glob pattern above it.
- Add last_used_at column to provider_connections CREATE TABLE schema
- Add ensureProviderConnectionsColumns migration for existing databases
- Add last_used_at to INSERT and UPDATE SQL in providers.ts
- Add last_used_at to JSON migration INSERT in core.ts
- Add zod to serverExternalPackages in next.config.mjs (#217)
Fixes#218: Round-robin routing strategy now correctly persists
the lastUsedAt timestamp, allowing rotation between accounts.
Fixes#217: zod module is now properly included in standalone/Docker
builds by declaring it as a server external package.
- Skip tools with defer_loading=true when assigning cache_control
(Anthropic API rejects the combination with 400)
- Delete stale schemas.js that shadowed the .ts source, causing
missing cloudSyncActionSchema export
Fixes#215
- Fix inverted Claude OAuth utilization (remaining, not used)
- Add hasUtilization() guard to prevent false exhaustion from empty responses
- Centralize anthropic-version into CLAUDE_CONFIG.apiVersion
- Add parseDate() for safe date validation in quota cache
- Batch background refresh with MAX_CONCURRENT_REFRESHES=5
- Move setModelUnavailable to after all accounts exhausted, not first 429
- Extract safePercentage() to shared utils (dedup)
- Use isRecord() type guard in usage API route
- Exclude binary files from Tailwind v4 source scanning
- Move setModelUnavailable from per-account loop to all-accounts-exhausted path
- Clear model unavailability on successful fallback
- Add in-memory quota cache with background refresh (5min active, 20min exhausted)
- Integrate quota cache in account selection to skip exhausted accounts
- Mark accounts as exhausted from 429 when no cached quota data exists
- Populate quota cache from dashboard usage endpoint
Merged all changes from the features-agente-mcp-a2a branch into a single
v2.0.0 release entry covering: MCP multi-transport (3 modes), 16 MCP tools,
A2A protocol, Auto-Combo engine, VS Code extension, consolidated Endpoints
dashboard with service toggles, 30-language i18n, full type safety overhaul,
and 1000+ tests.
- Created httpTransport.ts with singleton MCP server and WebStandard
Streamable HTTP transport running inside Next.js process
- Added /api/mcp/sse route (GET+POST) for SSE transport
- Added /api/mcp/stream route (GET+POST+DELETE) for Streamable HTTP
- Added mcpTransport enum to settingsSchemas (stdio|sse|streamable-http)
- Updated /api/mcp/status to report HTTP transport state
- Added TransportSelector UI with mode buttons and connection URL display
- Routes guard against disabled MCP or wrong transport mode
- Created settingsSchemas.ts with updateSettingsSchema to avoid webpack
tree-shaking the schema from the 908-line schemas.ts barrel file
- Updated settings/route.ts import to use new dedicated file
- Fixed focusRingColor lint error in ServiceToggle
- Added mcpEnabled/a2aEnabled boolean fields to updateSettingsSchema
- Rewrote ServiceToggle as clickable on/off switch with status indicator
- Toggle persists state via PATCH /api/settings
- Both services default to disabled (OFF)
- Renamed sidebar 'Endpoint' to 'Endpoints', removed standalone MCP/A2A entries
- Created tabbed layout with SegmentedControl: Endpoint Proxy | MCP | A2A | API Endpoints
- Added inline ServiceToggle (online/offline status) for MCP and A2A tabs
- Created ApiEndpointsTab placeholder with Coming Soon badge
- Updated i18n in en.json and pt.json with new endpoints namespace
Extracted validateBody, isValidationFailure, and loginSchema from the
935-line schemas.ts barrel file into a dedicated helpers.ts module.
Updated 70 API route files to import directly from helpers.ts.
Root cause: webpack on certain environments fails to resolve exports
from the bottom of large barrel files (schemas.ts), causing
'(0, O.Jb) is not a function' errors in production builds.
Fix: Split validation helpers into a small dedicated module (helpers.ts)
so webpack can correctly resolve all exports regardless of file size.
- TypeScript compiles with 0 errors
- All API routes updated to import from helpers.ts
- schemas.ts re-exports from helpers.ts for backward compatibility
- Issue #199: proxy_ prefix on tool names is now automatically disabled
when routing to non-Claude backends (Gemini, Compatible, etc.).
Prevents tool name mismatches in OpenCode, Cursor, and other clients.
- Issue #198: Added customRpm/customTpm support per provider connection.
Users can configure custom rate limits via connection settings,
overriding the default auto-learned limits from response headers.
- Fix createQuotaObject: utilization is percentage USED, not remaining
- Add optional chaining to window access in createQuotaObject
- Use typeof object guards for five_hour/seven_day instead of !== undefined
- Use nullish coalescing for extra_usage
- Propagate remainingPercentage in parseQuotaData claude case
- Use parseResetTime() for resetAt fields (consistency with other fetchers)
- Extract createQuotaObject() helper to reduce duplication
- Remove unnecessary Content-Type header from GET requests
- Use CLAUDE_CONFIG.usageUrl with template interpolation in legacy fallback
Eliminate `any` usage across the codebase by introducing proper generics,
typed interfaces (StatementLike, DbLike, PromptRow, etc.), and helper
conversion functions (toNumber, toString, parseVariables). Add
comprehensive Zod validation schemas for API endpoint inputs to enforce
runtime type safety alongside compile-time checks.
- Add type annotations to all providerModels helper functions
- Introduce LegacyProvider interface in providerRegistry
- Refactor translator system to use self-registering module pattern
with bootstrapTranslatorRegistry() and per-file imports
- Simplify translator/index.ts by delegating to modular translators
- Remove hardcoded Gemini OAuth client secret for security
Introduce full AI orchestration ecosystem:
- MCP Server with 16 tools, scoped auth, and audit logging
- A2A v0.3 server with JSON-RPC 2.0, SSE streaming, and task manager
- Auto-Combo engine with 6-factor scoring and self-healing
- VS Code extension with smart dispatch and budget tracking
- Harden CI pipeline: add static checks, remove continue-on-error
- Add translator schema validation tests
- Update .gitignore and CHANGELOG for release checklist
The OAuth usage endpoint returns utilization as percentage remaining,
not percentage used. Claude.ai showing 10% used corresponded to
utilization=90 from the API.
The Limits page showed "error" 0% for Claude Code (OAuth) providers
because getClaudeUsage() called /v1/settings which requires API key
with org admin access — unavailable to consumer OAuth tokens.
Now uses https://api.anthropic.com/api/oauth/usage with the
anthropic-beta: oauth-2025-04-20 header, which returns five_hour and
seven_day utilization data for OAuth accounts.
Falls back to legacy /v1/settings endpoint for API key users.
Introduce `normalizeQuotaResponse` and `normalizeCombosResponse` helpers
to handle varying API response shapes (array vs wrapped object)
consistently across MCP server and advanced tools. Add optional `meta`
field to checkQuotaOutput schema and update sourceEndpoints to reflect
current API routes.
- Add unit tests for 8 MCP advanced tool handlers (Phase 3)
- Migrate test files from JavaScript to TypeScript (.ts/.tsx)
- Restructure file paths from app/ to src/app/ across all tests
- Refactor route assertions into reusable assertRouteMethods helper
- Add tests for new API routes (compliance, audit-log, evals/[suiteId])
- Update barrel export tests to use consolidated assertion pattern
Register 8 new advanced MCP tools (simulate_route, set_budget_guard,
set_resilience_profile, test_combo, get_provider_metrics,
best_combo_for_task, explain_route, get_session_snapshot) with their
handler implementations. Add A2A smart routing skill that routes
prompts through the OmniRoute pipeline with routing explanation,
cost envelope, and resilience trace metadata.
benzntech