LATEST NEWS, tips and tricks!? What is the latest with Archon?

[Bosmanland]

Hey Guys,

I hope everyone is well! Wishing a healthy and prosperous 2026 to all!

I am reaching out to see if I am missing something!? I have not seen any updates or even discussions regarding Archon for a long time!?

Are people still using Archon? Am I the only one?

Are there any new tricks and implementations that anyone has to share?
Has anyone managed to integrate it with Google Anti-Gravity?
What is the latest Tech Stack users have that ARE using Archon?

Any news is good news in this discussion!

@coleam00

Cheers!

[alainivars]

Hi @Bosmanland , @coleam00 and everyone,

First let me give you @coleam00 and all the team behind you a BIG THANKS for your videos, they are great, and very instructives.

I asked myself the same think, I discover Archon just 2 months ago and digging and digging to understand everything inside and around, using it actively to store the data of my MCPs and I was thinking to contribute, but something at the moment is stopping me, who are the core engineer of the project (bug triage and pull requests review). That look like these persons are overloaded at the moment but forgot to delegate to some other one in the team the role for at list the time they are very busy.

So at the moment:

• Archon is supposed to be in version 2 beta, but the only tag that exist is : v0.1.0
• last merge in main (3 months ago)
• 105 issues open
• 45 pull requests are waiting

All that is not a good sign, but lot of people use Archon and the 13k+ like are here to prove it to you if you have any doubt.
You are the project leader and creator of this project, I do not forget the 31 other contributors + certainly some new one in the waiting pull requests. To all of you I say also a BIG THANKS, you made a great job, please continu, in all case you will be for ever in the hall of fame for your contribution to this beautiful project.

So what now, If all of you want this beautiful project to be used by a lot of professional like Opencode and other,
you should first add some policies like:

• the core maintainers should organize between them that the code and the documentation stay up to date and synchronized.
• any security issue is done first and urgently.
• any issue should have a description on how to reproduce it (except random bug, in that case the log should be provided)
• an issue recognized get a priority, and when some one want to take it, is assign himself to the issue, he/she become responsable to provide a fix.
• a pull request that comply to the quality should be review, accepted and merged ASAP
• a pull request that modify something describe in the documentation MUST also contain the update of the documentation.
• after every merge the core member set a new tag.
• the policy for the tag is: bug fix increment the last digit of the version., new functionality incorporated the middle digit of the version is incremented, major if that update make the server not compatible with the previous version.
• Claude and other should not be able to commit in they name, they did't worked alone, some one have to take the responsibility of the commit.
• Never forget to have a release note.

You should also create some task to update the documentation, and add in the README.md the description to launch archon-docs container, it took me 2 months and a deep reverse code to discover that this nice documentation exist, don't be a shame, that documentation is well done. Please also add the endpoint documentation for FastAPI at http://0.0.0.0:8181/docs and the documentation for all the MCP endpoints.

To finish, at the moment I actively looking for my next contract, so that mean I work full time of Archon because I like it, I even just discovered today that in Claude (I don't use it) or Opencode you can after install an hidden documented part of Archon, you can type /archon-onbording and the AI guide you to fix you first issue in Archon and that is a genial idea, that the first open source project I see with this functionality.

Thank's all of you again for this beautiful project, do what ever you want with my advice's but, please don't let die this project, if you need some hands, let me know, I already contributed to some other open source projects.
A.I.

PS: no AI was used to redact this message, it is a message from human to human.

[Zebastjan]

Hi @alainivars,

You’re absolutely right to notice the tension between how widely Archon is being used and how “stalled” things can look from the outside when merges, tags, and issue triage don’t keep pace with community activity. The numbers you pointed out (open issues, PR backlog, old tags) are real signals, and your proposed policies are pointed in exactly the right direction.

A few things I wanted to share from our side:

There is heavy development happening on Archon right now, and we hope it will be merged into the main project. A lot of it is focused on getting the foundations right rather than small surface changes. We’ve been working on a more robust ingestion pipeline (checkpointing, resume after crash, sanity checks on ingested data, A/B testing of summaries/embeddings, etc.), because so many of the “higher‑level” features depend on trustworthy data going into the system.
​

In parallel, we’ve started to formalize a chunking roadmap for the RAG side: moving from mostly size‑based chunks toward structure‑aware and eventually semantic chunking (paragraph/heading‑aware chunks, Docling‑based document structure, better handling for code, and later semantic/embedding‑driven boundaries). This ties into future work on better summaries, tagging, and retrieval quality.
​

We’re also sketching out clearer versioning and backend abstraction plans: how to handle multiple dataset revisions, potential alternative vector backends, and eventually graph/relationship views.
​

So while the main branch and tags may not fully reflect it yet, there is active design and implementation work happening around:

Robust ingestion and recovery.
​

Smarter, configurable chunking for both docs and code.
​

A clearer long‑term roadmap (including IPFS/shared KBs, database abstraction, and a knowledge graph layer).
​

On the community/process side: we agree with a lot of what you’re suggesting around clearer maintainer roles, issue hygiene, PR review discipline, tagging/versioning policies, and making the docs (and FastAPI/MCP endpoints) much more discoverable. Those are exactly the kinds of “operational maturity” steps we want to grow into as the project stabilises around its next major iteration.

And yes – contributions like the ones you’re thinking about (especially in areas like chunking, ingestion, and documentation clarity) are very welcome, and we’d absolutely like to see them merged once they align with the evolving architecture. The fact that you’ve been using Archon daily, reverse‑engineering parts of it, and still care enough to write this kind of message says a lot about the value you’d bring as a contributor.

Thank you again for the encouragement, the concrete suggestions, and the time you’ve already invested in understanding and using Archon. Please don’t hesitate to open issues/PRs or propose concrete changes – we’re actively working toward a place where those can be reviewed and integrated more quickly and predictably.

Best,
Sebastjan

[alainivars]

Thanks, for your answer @Zebastjan.

[quarantaquatro]

Hey everyone,

It's great to hear the project is still alive! However, it's been a few months since the last update (around 4-5 months, I think?). With so many new tools popping up all the time, this might make Archon look a bit behind. I still really like Archon, but things are moving so fast it's getting tough to stick with something that hasn't seen much recent activity. Huge thanks to the whole team who have been working on this, and those who are still developing! It would be awesome if the original team could also contribute with some updates and tackle those pull requests - it would really boost confidence in the project's future.

[lfontanez]

Hola!

I use Archon every day as main MCP server for my Aider-Desk + local-ai-packaged setup. The ingestion pipeline is excellent, the RAG works well, and the task management features are solid for centralized use.

It fills an important gap in my workflow and clearly helps my AI agents produce better results. There’s still room for improvement—better handling of 429 errors and clearer versioning would be great—but it’s in beta, and it’s already incredibly useful.

Please keep cooking!
I will review the contribution process and see if I can help.