engomez

Closing loop - we merged in a PR that lifted the codex toml modification logic from the codex source code, so it should be safe to use. We weren't able to recreate cases were the previous approach broke our test cases, but by porting codex logic we figure is the best way to structurally ensure there aren't unforseen changes.

If you run into any other issues -- let us know.

Added! https://openknowledge.ai/docs/integrations/opencode

OpenCode docs added - https://openknowledge.ai/docs/integrations/opencode

Let us know if any feedback.

Top priority, expect an update within days.

Appreciate it !

Currently yes. We'll be exposing a centralized CRDT server option in near future.

Yes, we're working on Linux/Windows beyond the CLI/WebUI, Electron enables us to do that.

Yup, OKF is more of a specification for a format, similar to e.g. Skills or MCP.

Our templates follow it and we have a starter pack for it.

You can think of our app as a general purpose IDE for any knowledge base/wiki/markdown.

Docs: https://openknowledge.ai/docs/workflows/supporting-open-know...

Ah, I was referering to https://cloud.google.com/blog/products/data-analytics/how-th... for "OKF"

Talking about a few different things, OKFN and OKF.

Web UI available for Linux/Windows. Updates docs !

I don't believe I've seen a KM app with true WYSWIG that's markdown underneath. Rest is indeed mainly packaging up all the second-brain, MCP, skills, and AI integrations to be out-of-the-box. More to come.

Fixed!

As a founder of a company that revolves around AI, I get you. It can feel frothy // intangible // overwhelming.

There is a knowledge graph -- we'll update the readme to make clearer. We support wikilinks and forward/backward links are presented to the agent through the MCP server for "agentic search". There's semantic search support as well, etc.

Great thought - we indeed added Claude/Codex terminal in there already. Was our quick win as we investigated this deeply.

collab between agent <> human happens locally via yjs

collab between human <> human happens using git/GitHub. so collab right now is auto-sync (~few min latency) + sharing functionality. Richer realtime collab is indeed something we're looking at.

Thanks nitin, these are all top of queue. We'll follow on quickly.

Appreciate it. An anonymized repro file would help ensure we validate against your specific cases.

got it. we'll ship this within the next week.

To add one more thing: Codex/Claude/Cursor can open the OpenKnowledge web viewer within their own embedded web viewer. So you don't need the two apps open.

And we embedded the Claude terminal within the OpenKnowledge app itself if you prefer that. We are working on embedding the AI (including local models) more deeply within the app itself as well, expect updates in next week.

Is the issue you're seeing that it stripped comments, or something else?

Got it yup, sounds like similar to the setups our team runs. Give it a shot and feel free to share any feedback.

Git is used for the "auto-sync" and sharing functionality. Uses GitHub as the source of truth for the content.

Currently the CRDT is local and is used so that agents can edit the markdown concurrently with the user, and the user can edit it via the WYSWIG editor or the raw markdown editor. CRDT powers the live indicators, etc.

CRDT and git are reconciled so that git stays as the canotical version history.

Let me know if you see any gap with OpenKnowledge an this scenario. It's exactly what we aimed to solve (files are markdown and local, sync and team sharing happens via git/GitHub but is abstracted away from user).

Checking this asap, apologies for any inconvenience.

Curious: do you mean being able to select text and "add it" to clipboard/attachment for the LLM, or something else?

+1. We'll continue at it.