OneSearch v1.4.0: a self-hosted search layer for files across your machines

OneSearch v1.4.0 released yesterday and it can now search files that live on other machines, not just what's mounted on the server. OneSearch is a self-hosted search layer for files you already have. It indexes folders in place and keeps them searchable from a web UI or CLI. The intended flow is pretty boring: mount folder -> add source -> index -> search The big v1.4 change: remote indexing agents . On my very first post here , someone asked whether they could index files on separate machines back to one central instance without mounting everything onto the server. It stuck with me. I'd been sketching it out for a while and the build itself took about 2 months lol. Probably the biggest feature I've added since launch, learned a lot through it. xd You install a small agent on the other machine, enroll it, approve it from the admin console and add directories from its allowed roots as sources. The agent connects outbound to your server, so theres no inbound port to open on the remote box. There are native builds for Linux and Windows and a Docker image. Agent updates are signed too. Already-indexed content stays searchable when that machine is offline. You just need the agent back online to grab an original. Ofcourse, using indexing agents is optional and local mounting works as before. Agentic OneSearch? jk. old school agents... --- How I see the overlap with other apps: Paperless/Docspell are better for document intake, OCR, and archival workflows. Nextcloud search is better if your files already live in Nextcloud. Recoll is great for single-machine desktop search. OpenSearch/Fess-style setups are better if you want a big configurable platform. --- OneSearch is for existing-file search across NAS shares, bind mounts, old project dirs, manuals, ebooks, media folders, the stuff you don't want to move into a new workflow. Those folders can now live on separate machines too. The stack is FastAPI, React/TypeScript, Meilisearch, Docker and a Python CLI. If anyone knows that stack and wants to poke at it, contributions or technical feedback would be welcome, esp around deployment testing, frontend cleanup, file extraction edge cases and indexing behavior. Have gotten some really useful feedback and discovered edges case from community before, so I am really grateful for that. Repo | Docs | v1.4.0 release notes --- Also curious how people here solve existing-file search today. Recoll, Nextcloud search, Paperless, OpenSearch/Fess, custom scripts, something else?