Thefux

Bridge
keep everything local, but make it accessible from anywhere through open infrastructure.
CEP-22 and CEP-41 forced me to start playing with @nprofile…g2s8.
So I tested with something I work with everyday, my hermes agent. An experiment that I Think it was worth every sat paid on @nprofile…gs7e to build it.
Sending media files over ContextVM is probably not what it was designed for, but it's a discussion for a different post.
hermexVM
@nprofile…gs7e for open model routing.
@nprofile…g2s8 over Nostr for the encrypted connection.
What you can do in a conversation
● Chat with any configured Hermes agent and receive responses live.
● See thinking, status updates, tool calls, progress, output, and errors as they happen.
● Approve or reject sensitive tool actions and answer the agent’s clarification questions.
● Reply to or copy individual messages.
● Render rich Markdown, including code blocks and links.
● Take Photos, attach images, audio, video, PDFs, archives, code, text, and other files.
● Record voice messages and turn them into an editable transcription before sending.
● Send selected messages or a full transcript to another agent, either as a new conversation or into an existing one.
Features
● Multiple Hermes agent profiles.
● Persistent conversation history with automatic resume and live reattachment after connection drops.
● Model selection for each conversation.
● Project selection and project creation.
● Searchable Hermes skill picker, grouped by category.
● Linked conversation and agent-handoff history.
● Automatic upload retry and recovery from temporary relay failures.
● Explicit client allowlisting at the bridge.
● Android and web support—the web client can also be built and published as a Nostr nsite.
● Connect to multiple Bridges.
https://blossom.jumble.social/5ee82c780284feb4ed0f20006350023e34c474d5d71c537cfce73399822b4133.png
https://github.com/9qeklajc/hermexVM
and on @nprofile…gsu4
https://zapstore.dev/apps/ai.hermex.vm
#Bridge #hermexVM #Hermes #Routstr #ContextVM #Nostr #localfirst

2026-08-23 13:56:06 CEST

- reply

4.5 is not released yet, that's probably why.. We've been fixing some of the issues lately, minitbits is likely due to ratelimits which was optimized lately too .. We are releasing soon

2026-07-21 22:15:54 CEST

- reply

If u mean bloom then u could change the port or even disable the the relay .. At least in the version I'm running in my phone rn

2026-07-21 22:13:26 CEST

- reply

Which app u mean

2026-07-06 21:25:26 CEST

- reply

Same model will not show up twice for two different provider .. The requests will be always forwarded to the first ("cheapest") upstream and fallback to the other if failed ..

2026-07-06 16:03:12 CEST

- reply

Thanks for the details, will look into it and keep u updated ..

2026-04-14 18:50:19 CEST

- reply

The model structure changed slightly that's why the failed to parse message ... I just fixed it in the main branch, we will be releasing soon ..

2026-01-25 00:07:41 CET

- reply

The app will have both Blossom and Relay, will make some more tests .. hope by next week on zapstore

2026-01-05 23:27:41 CET

- reply

Not sure if I get u right .. I just didn't want to reinvent the weehl for my use case, let's have a more concrete example of what I want to do, upload a presentation (.mp4) and link the slides (.PDF)s to it. I get ur point about the protocol and app part, thought maybe there is already something done for the same thing that I'm proposing ..
I'd probably introduce a new event kind to preserve that relationship between one or more file types (blobs)
What do u think ?

2026-01-03 20:49:35 CET

@nprofile…evew is there an option or a way to make a blob reference multiple blobs, use case would be for example creating some ai summaries/notes or just add images (for MD files) .. Probably NIP-94 could be extended I think or is there any approach already discussed here I'm not aware of ? Just got the idea while planning to create an app based on BUD-11

2025-10-14 16:42:50 CEST

- reply

Routstr also supports the x-cashu (NUT-24) header protocol .. u don't need to have any balance by ur provider, basically u pay per request .. u will need to have a client that can extract the model cost per request and can generate a cashu token .. https://otrta.routstr.com/documentation/overview
Other projects are following
https://njump.me/nevent1qqsfqhkmy50fw6l3kw3c6pr9f8pzpc3r36ce44wz3awzkflcczs26wcppemhxue69uhkummn9ekx7mp0qgsthdwa5rs42euhnuz5xsrmmssr84hshwes7uj392vpeldj7z0zw3crqsqqqqqpn7fhyt

2025-10-09 16:50:26 CEST

- reply

Yeah that would work, but is it an issue to let the relay do that by itself .. With contexvm, we need probably a second service that is somehow connected to the relay to apply the changes .. @nprofile…e5wx correct me if I'm wrong

2025-10-09 16:23:38 CEST

@nprofile…278z @nprofile…ufa8 @nprofile…6htm any though on using nostr events to store relay configuration, like white listed npubs, etc.. , relays could fetch the event and apply on the fly without a restart..
It could also be a public event so anyone can copy..
Thought myself of having a relay daschboard to update settings and probably use it for other things..
(Or do u know about anything like this?)

2025-10-08 14:42:46 CEST

take charge with Wally: a sleek, next-generation wallet that centralizes your finances and even monitors your LLM usage and setup a badget, made by @nprofile…6htm @nprofile…08mz . Its plugin architecture lets you tailor the experience to fit exactly what u need .
#SovEng
#Otrta configure ur nwc connection and topup ur balance.. imagine the topup runs automatically 🔥 (For now, top-ups need to be done manually.)
https://blossom.primal.net/b1ca3aa94fda59d570c7b37ccc06128547cae9ed27b828855d6c11f7cd1674cf.mov
#Routstr configure ur routstr proxy @nprofile…nq5z
https://blossom.primal.net/31fbd817b3c6df085effccaa5df83bc024ef37115be89ff993278ac535ae7eaa.mov

2025-10-08 12:00:37 CEST

- reply

soon 🫡
https://blossom.primal.net/47f0b8fbf69c83e1e169d1b63f48a10176360f091a925cda79c637f93a12bf6d.mov

2025-10-02 02:46:49 CEST

- reply

What if the event is send to all configured relays and the first one who redeems it gets it, so its like mining the event bit still all have to keep it .. So relays could compete 😸

2025-09-08 22:12:22 CEST

otrta 0.1.2
We're thrilled to announce some powerful new features that enhance your AI experience:
* Nostr Marketplace Discovery: Easily discover providers with NIP-91 support!
https://imgproxy.coracle.social/x/s:640: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
* Model Comparison: Find the best model prices with our new comparison view!
https://imgproxy.coracle.social/x/s:640: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
Future Plans:
We're already working on more innovations, including:
* Automatic provider switching for the best model prices!
* setup api key budget limit
Stay tuned for more updates!

2025-08-22 21:38:52 CEST

- reply

Users are not required to have account/balance by their provider they could switch provider instantly without asking for refund..
when user sends a request an ecash token is created and added to the request, providers forward the request to an ai provider (like openai) get paid with the token provided and return change (if any) .. Basically the user has always a new identity

2025-08-10 09:25:03 CEST

- reply

Hey, are u still having issues, currently the live otrta is up and running .. Not sure but could it be u logged in with different npubs ?
PN if the issue persisted

2025-08-08 19:14:05 CEST

- reply

It's 'one to rule the all', we still look for a cool pronunciation 😅

2025-08-08 00:35:11 CEST

- reply

U could use zed with routstr client already .. all u need is to setup the provider, basically ucould do that here: https://otrta.routstr.com
Checkout the dashboard when u login u will find the URL u need (create the api key, select ur provider, topup ecash to the provder mint and ur good to go)
https://zed.dev/docs/ai/llm-providers#openai-api-compatible

2025-08-07 11:34:07 CEST

- reply

Did some work and summarized all in this project..
https://njump.me/nevent1qqspjv7e6k0yka9sqrgfkrxphgrlkqpcur9zk2qat067998ghu8gcuqpz4mhxue69uhkummnw3ezummcw3ezuer9wchsygpzyndp04mthpllgcg7au3evtdqfu254xa7vsf4tl9kgj8rt8fzpgpsgqqqqqqsyedmvv

2025-08-07 08:40:03 CEST

from Project for a friend to otrta routstr client the Privacy-First AI Access!
We're thrilled to announce **otrta** - the revolutionary Routstr client that brings you completely anonymous AI access powered by cutting-edge e-cash technology! 🚀 With private micropayments using Cashu e-cash notes that let you pay for exactly what you consume down to the millisatoshi. Whether you're a solo developer or managing a larger team, otrta seamlessly integrates with your favorite AI coding tools like goose, Cline, Roo Code, and Kilo Code - just update your API endpoint and you're ready to go!
**Complete Control & Privacy**: Built for ultimate privacy and autonomy, otrta is fully **self-hostable** - deploy your own instance and maintain complete control over your data and infrastructure. Connect to AI providers through both **standard URLs and Tor hidden services** for maximum anonymity and censorship resistance. Built on the decentralized Routstr protocol with Nostr authentication, you get complete privacy, flexible deployment options.
Join the privacy-first AI revolution and experience what truly permissionless AI access feels like - no KYC, no credit cards, just pure innovation at your fingertips! ⚡
📋 **Check our release notes** → [Release 0.1.0](https://github.com/Routstr/otrta-client/releases)
📚 **Explore comprehensive docs** → [Documentation](https://otrta.routstr.com/documentation/overview)
🚀 **Start your journey now** → [Login & Join the Revolution](https://otrta.routstr.com/)
## 🙏 Special Thanks
@nprofile…0e3k - Our relentless testing champion who battled through countless iterations, found every edge case, and pushed the boundaries until we achieved bulletproof reliability.
@nprofile…xnfk - The e-cash wizard whose expertise shaped the core of our payment infrastructure.
#routstr #ai

2025-07-28 11:52:23 CEST

- reply

We are planning to release this week .. pre-release https://otrta.routstr.com

2025-07-21 22:05:03 CEST

- reply

hey Arjen, thanks for reaching out .. That part of the documentation is still missing but we are currently working on the specification.. the current implementation of routstr does accept ecash from any mint so probably there would be no issue with using any mint proofs u have ..
reach out if u face any issues

2025-07-13 01:27:54 CEST

- reply

Sure, I mean to make sure ur server is reachable u could do this simple test
curl -X GET https://embeekttrsdquyyicngh5i6xc6xkqymm6cihgzngtbxpoevd6ut5pead.onion/v1/models
That should return the list of all models in ur models.json file
u could also get the the server logs. First navigate to the folderwhere ur docker compose file is located and run the followingcommand
$ docker compose logs

2025-07-13 00:48:48 CEST

- reply

Nice !!
One think to make sure ur server can be reached, try to collect ur models.
a GET request to this endpoint '/v1/models'

2025-06-26 19:04:29 CEST

- reply

Yeah I'm currently cleaning up and improving some stuff, but basically the code is about the client that the user runs locally and hook up to any ai driven software that use the openai api protocol (like. Goose, roocode, cline, etc.).
will soon release the client with more docs and features..
Ppq or other provider should only implement the interface based on the provider-cliemt spec which as u mentioned still not well documented yet (working on it 😉)

2025-06-26 00:04:10 CEST

- reply

I joined routstr team recently.
Did myself worked on such solution
https://github.com/9qeklajc/ecash-402-client
Where u pay ecash per request.
This method ist for sure as u mentioned it, not perfekt, but its a decision the provider make and risk loses ..
Wrote also some ideas on the topic, about a smart client and dynamic provider.
https://github.com/ecash-402/ecash-402-specs
Feedback is welcome

2025-06-16 23:03:39 CEST

- reply

Not realy familiar with openrouter but, have you tried litellm ? Would you recommend it ?

2025-06-11 08:22:32 CEST

- reply

To prototype and to see if ideas make sense then yeah vibe coding is awesome, but fortunately the currently available tools and llms generate a lot of noise after multiple iterations (e.g unused code, inconsistent style, etc..) and that makes it more harder to maintain and for others to contribute

2025-05-17 13:32:34 CEST

- reply

30 SATs per request are minimum

2025-05-17 13:29:44 CEST

the api could be easily tested via curl
curl -i -X POST https://ecash.otrta.io/v1/chat/completions \
-H "X-PAYMENT-SATS: <ecash token>" \
-H "Content-Type: application/json" \
-d '{ "model": "meta-llama/llama-4-maverick-17b-128e-instruct-fp8", "messages": [{"role": "user", "content": "Tell me the difference between ecash and regular money."}] }'
The response includes a new header field (X-CHANGE-SATS) that provides an eCash token with change.
x-change-sats: <cashback token>

2025-05-16 17:28:38 CEST

- reply

That would be the easy part (I think), when I setup my own mint ..

2025-05-16 12:22:26 CEST

- reply

Saw it but didn't really have the knowledge and capacity to setup all those technologies together ..
Was really happy when I found nutshell to use via API (currently the API ist not supported anymore 🙈) so an older version is used!
There still a lot to do ..

2025-05-16 08:29:46 CEST

Project for a friend 😁
Excited to announce the ECASH OpenAI API client! There's still a lot to be done, contributions are welcome 🙏
https://github.com/9qeklajc/ecash-402-client
https://cdn.nostrcheck.me/2224da17d76bb87ff4611eef23962da04f154a9bbe641355fcb6448e359d220a/f8538e7cef2095be7335bfcde91411320daf1555ac1fed559fc3c6f8be0e9d96.mp4