Security:
- Fixed unauthenticated SSRF in /api/stream/proxy and /api/stream/ts-proxy
- proxy allowlist had escape hatch via pathname.endsWith('.m3u8') bypass
- ts-proxy had no destination validation at all
- Both could fetch internal services (127.0.0.1, 169.254.169.254)
- Added streamProxy.helper.js with domain allowlist, private IP blocking,
and http(s)-only enforcement applied to both endpoints
- Removed dead megaplay-1.buzz; added vid-tube.site, norami.top,
kryntal.top, akirax.buzz to allowlist
- Added SECURITY.md with vulnerability reporting policy
Bug fix:
- Fixed dotenv.config() loading after static imports -- helpers reading
process.env at top level never received .env values. Changed to
import 'dotenv/config' as first import
Config:
- Added STREAM_PROXY_DOMAINS env var for custom allowlist
- Added STREAM_PROXY_REFERER env var for custom CDN Referer
Root cause: upstream /ajax/server and /ajax/server/list endpoints require
session cookies set by visiting the watch page. fetchWithMirror() didn't
persist cookies between requests, causing upstream to return 500.
- Extract cookies from set-cookie headers when fetching watch/home page
- Pass cookies with AJAX requests for stream info and server list
- Accept optional slug query param in /api/stream for targeted session
- Both extractors now use cookie-aware axios for session establishment
- Fix extractServerList JSON parsing: fetchWithMirror returns raw text but code treated it as parsed object, causing Cheerio to parse JSON string as HTML resulting in empty server arrays
- Fix extractMapperServers JSON parsing: same raw text issue caused data processing loop to be skipped, returning empty arrays
- Fix streamInfo controller to return 404 with error message when stream URL is null instead of silently returning null values
- Fix server list controller to use proper TTL and handle empty results
- Fix JSON.parse error handling in episodeList and episodeListAjax extractors
- Update CHANGELOG.md with v2.2.1 bug fixes
robots.txt: Allow / (landing page), disallow /api (JSON endpoints)
sitemap.xml: Updated lastmod to 2026-07-28, removed non-HTML URLs
index.html: Updated meta tags with 38 endpoints, richer keywords,
JSON-LD schema.org/WebAPI with accurate metadata
manifest.json: Updated theme/background colors to match site (#faf9f5),
better description with 38 endpoints
Key SEO changes:
- Title now includes '38 Endpoints'
- Meta description expanded with keywords (hls proxy, m3u8 proxy, etc)
- JSON-LD provider type changed from Organization to Person (author)
- Added license field to JSON-LD
New endpoints:
- GET /api/stream/proxy?url= — M3U8 playlist proxy with URL rewriting
Rewrites relative .m3u8 and .ts URLs to proxy endpoints
Adds CORS headers for cross-origin playback
Validates allowed streaming domains
- GET /api/stream/ts-proxy?url= — TS video segment proxy
Proxies .ts segments with proper Content-Type and caching
Fixes:
- seasons.extractor.js: Now extracts from watch page sidebar instead of broken AJAX endpoint
- watchOrder.extractor.js: Now extracts from watch page sidebar (trending/related)
- mirror.helper.js: 404 responses no longer mark mirrors as failed
Only connection errors trigger mirror failover
Endpoint-specific 404s break immediately without retry
38 total endpoints, 34/34 passing with 0 errors
New endpoints:
- GET /api/stream/resolve — Resolves actual m3u8/mp4 URLs from embed player URLs
Follows chain: vidtube.site → megaplay API → m3u8 playlist
Returns url, type, qualities, subtitles, skipData
- GET /api/stream/qualities — Parses M3U8 playlists for available quality options
Returns sorted quality list with resolution, bandwidth, URLs
New extractors:
- streamResolver.extractor.js — Full stream resolution pipeline
- resolveStreamUrl(): Resolves embed → actual video URL
- parseM3u8Qualities(): Parses master playlists for qualities
- normalizeServerName(): Maps VidPlay-1 → vidplay, HD-1 → hd
- extractSubtitles(): Extracts subtitle tracks from API responses
Enhancements:
- streamInfo.extractor.js: Added session establishment before AJAX calls
- Server list now includes normalizedName field for each server
- Stream info now includes type (hls/mp4) and backup URL
- Server list includes typeLabel (SUB/HSUB/DUB)
- Session cookies properly maintained for AJAX stream endpoints
Architecture:
- Embed API domain mapping (vidtube.site → megaplay-1.buzz)
- M3U8 parser supports master and media playlists
- Subtitle extraction from multiple API response formats
- 36 total endpoints (was 34)
New Endpoints (4 new routes):
- GET /api/upcoming — upcoming anime releases from homepage
- GET /api/top-rankings?sort=top|newest — ranked anime with sort modes
- GET /api/recently-updated?tab=all|dub|sub — filtered recent updates
- GET /api/completed — finished anime series from homepage
Enhanced Filter (/api/filter):
- Added source parameter (18 source material types)
- Added epMin/epMax for episode range filtering
- Added excludeWatchlist parameter for logged-in users
- Added season parameter (spring/summer/fall/winter)
New Config IDs:
- SOURCE_IDS: 18 source material type mappings
- SEASON_IDS: 4 broadcast season mappings
- EXTENDED_TYPE_IDS: Added tv-short and tv-special types
Total endpoints: 34 (was 30)
OpenAPI spec updated with new endpoints
CHANGELOG.md updated with comprehensive change history
Major Changes:
- Add shared parseListItem.helper.js for unified list parsing (DRY)
- Migrate all 22 extractors to use fetchWithMirror() for consistent mirror fallback
- Remove unused cors dependency (custom CORS middleware handles it)
- Fix middleware ordering: timeout now registered before routes
- Fix CORS to allow POST for /api/mirrors/reset endpoint
- Remove redundant cache check in homeInfo extractor
New Features:
- Request ID tracking (X-Request-Id header) for debugging
- Graceful shutdown handling (SIGTERM/SIGINT) with 10s timeout
- CSP and HSTS security headers
- Rate limiter cleanup interval (every 5 min) to prevent memory leak
- Request body size limits (10kb) via express.json()
Security:
- Content-Security-Policy header for XSS protection
- Strict-Transport-Security for HTTPS enforcement
- X-Permitted-Cross-Domain-Policies: none
- Body size limits to prevent large payload attacks
Code Quality:
- All files follow strict documentation style (box headers, section separators,
feature markers, JSDoc params/returns, inline notes, module footers)
- Eliminated ~200 lines of duplicated code across 8+ extractors
- Updated CHANGELOG.md with comprehensive change history
- Add pagination.helper.js with addPaginationMeta function
- Add pagination info to: most-popular, search, new-release, newly-added, latest-updated, az-list, filter, category, status
- Pagination includes: currentPage, totalPages, totalItems, itemsPerPage, hasNext, hasPrev
- Fix creatorInfo middleware to include message field
- Fix bug where full URLs were passed to fetchWithMirror causing doubled URLs
- Extract path from full URL before passing to mirror fallback
- Fixes search, most-popular, new-release, newly-added, and az-list endpoints
- Update endpoint count from 30 to 33 in badge, stats, and descriptions
- Update caching description from 5-min TTL to LRU with configurable TTL
- Update timeout description from 15s to configurable
- Update mermaid diagram labels
Browser blocks cross-origin requests from AniKotoWeb because
Access-Control-Allow-Origin was never set when ALLOWED_ORIGINS
was undefined on Vercel. Changed condition to allow all origins
when env var is not configured.
- apiRoutes.js: Moved imports outside comment block (caused 500 on all endpoints)
- test.js: Deleted orphaned duplicate code after function close
- random.controller.js: Removed stray closing brace breaking try/catch
- cache.helper.js: Added missing MAX_CACHE_SIZE constant
These bugs were introduced by sed insert commands placing code in wrong positions.
The API was returning 500 FUNCTION_INVOCATION_FAILED on all endpoints.
Shinei Nouzen