A self-hosted router that sits between your applications and every model provider.
One endpoint, three pass-through surfaces — OpenAI chat completions, Anthropic
messages, and OpenAI Responses — with namespace routing across multiple provider
accounts, failover combos, and encrypted key storage.
Its ledger records every call and normalises token accounting across all three
protocols, so cost and usage are attributable per account and per model instead of
scattered across provider dashboards. An embedded dashboard serves the live call
feed and budget controls.
Source ↗
A public web service serving Malaysian prayer times and the Qur'an in Malay and
English. Its own API and caching layer sit in front of the official JAKI timetable
source, and the same logic ships as cpray, an npm module other developers
can drop into their own projects.
waktusolat.me ↗
Source ↗
When standard indexers give up on a missing movie, show, or album, eziarr keeps
hunting: manual and automated tooling that searches alternative sources —
Telegram, the Internet Archive, open directories — inside your existing
*Arr stack.
Source ↗
Real-time insight into a Jellyfin server: who is streaming what right now, direct
play versus transcode, library statistics, and a user roster — built to stay light
where conventional monitoring tools get heavy.
Source ↗