Cameri
Cameri🐲

🤯🤯🤯
Code-first video as pure software engineering
Forget Photoshop, After Effects or Blender… with this skill you can just ask your coding agent
https://github.com/gatewai-dev/gitframes
Examples:
https://youtu.be/9BF1snfmg3o
https://youtu.be/FVXTqAozAv8
https://youtu.be/RQ3yi_lzPqI

2026-10-05 17:55:12 GMT

The “AI going rogue” to be inflicted by the regulate-me-harder bros is going to make 9/11 look like the boogeyman in comparison.

2026-10-04 16:16:06 GMT

- reply

But if it’s art, why don’t we see it on the walls, framed in photos, recited in plays, sold in antique stores, winning linting competitions, hung in museums, and the like?
Because code its not art. One can be proud of something you wrote because one has written it. I know I have been. Just like you can feel proud you put IKEA furniture together sans the manual.

2026-10-04 16:10:27 GMT

- reply

100%

2026-10-04 15:49:23 GMT

The “We want our code to be art” crowd is cringe AF

2026-10-04 03:40:24 GMT

- reply

Hmm worth exploring!

2026-10-03 15:11:29 GMT

- reply

Muse Home Link, which you’ll need to give Muse access on your network. Unless they ship a container image you can run…

2026-10-03 15:09:02 GMT

- reply

Interesting! Nope, but this looks like the thing I need for my Software Factory!
I’ve kind of been building my own thing, not using other people’s framework. So it’s stitched together with agent fleets that tan communicate internally with each other and externally with other fleets over NATS.
With Synadia’s Nex I’ve been building my own internal Lambda service which could come in handy for this.

2026-10-03 14:43:17 GMT

- reply

No thoughts yet, but I will spend some time noodling on it.
Docker came up with their “sandbox” feature but I haven’t tried it.

2026-10-03 14:13:28 GMT

- reply

I run them inside a sandboxed container.
The agent only has access to the paths on the host I give it permission.
The network egress is gated by a mitmproxy + wireguard setup and each remote host must be approved explicitly by me.
It also has Docker-in-Docker access with limited permission (mostly read) gated using Open Policy Agent (OPA).
Secrets are stored on disk encrypted using SOPS / AGE and always scoped.
If you want to set up your own use these two schematics:
https://github.com/cameri/schematics/blob/main/schematics/assemble-a-sandboxed-agent-set/SCHEMATIC.md
https://github.com/cameri/schematics/blob/main/schematics/add-an-agent-harness/SCHEMATIC.md
Currently it works with Docker, and I’m going to play around with Podman.

2026-10-03 13:19:43 GMT

- reply

US only

2026-10-03 13:00:50 GMT

Oh My Pi is great!
And with herdr it’s a killer combo.
https://omp.sh/

2026-10-03 12:57:43 GMT

Nostream v3.2 dropped!
https://github.com/cameri/nostream/releases/tag/v3.2.0
Recent updates:
Version 3.2 (October 2026)
- Relay owners can hide posts that trusted moderators reported
- Updates with no downtime for busy relays
- Fixed searches that mixed several tags
Version 3.1 (September 2026)
- Search inside posts
- Private messages show only to the person they belong to
- Invite-only relays with invite codes
- Spam protection that gets stricter when the relay is busy
- A dashboard and alerts for the person running the relay
Version 3.0 (June 2026)
- Reactions, subject lines on posts, and saved relay lists
- Post counts: a relay can say how many posts match a search
- Proof of date for a post (OpenTimestamps)
- Group chat with end-to-end encryption (Marmot)
- Relays can take payments through Nostr Wallet Connect
- Fixed a way to fake your network address
Many of these are off until the relay owner turns them on, so check the settings!

2026-10-03 12:55:49 GMT

GM Nostr!

2026-10-02 17:32:03 GMT

https://schemaformat.ai/

2026-09-30 12:27:29 GMT

cc All Nostr Philosophers
https://x.com/iampascio/status/2098799094835675574

2026-09-30 00:28:01 GMT

- reply

I should have thrown sats is what you are saying?

2026-09-29 23:19:13 GMT

Throwing tokens at my problems

2026-09-29 12:51:53 GMT

- reply

Jokes aside: that sounds like a problem that should be solved at the protocol level 🪄

2026-09-29 12:50:12 GMT

- reply

Nostr Hard Fork Incoming!

2026-09-09 01:16:46 GMT

That Little Snitch red warning had me 😱

2026-09-06 15:38:08 GMT

- reply

Yup, I can see it.
It reminds me of AR solutions for industrial plants where an operator could see an overlay of sensor data of physical equipment, among other things. Not sure where that tech is today.

2026-09-06 15:33:50 GMT

Anthropic models cost a magnitude (we are talking 20x plus) more than DeepSeek V4 Flash or z.AI’s GLM 5.3 Flash for the same tokens.
Most workloads don’t need Fable let alone Opus levels of intelligence, and these will refuse to help you when and where you truly need them like with security.
The 5-hour and weekly limits on Pro and Max plans will limit your ability to ship or cause you to overpay putting a cap on your efficiency. Eventually you’ll be forced to use their API rates which are the most expensive.
What most people miss is a Software Factory, a repeatable workflow that yields good enough reliable software you can spec, build, and ship that sizes and break downs the work into workable units.
You shouldn’t be building software but the machinery that does it for you.

2026-09-06 15:17:31 GMT

- reply

I hate those apps with passion.

2026-09-06 15:14:16 GMT

- reply

Nice! Comes in clutch cause I cancelled that one app I was using: Built With Science.
Does it have a “tough love” or “emotional empathy” personality? 🤨

2026-09-06 15:07:14 GMT

- reply

I should’ve known better… 🤣

2026-09-06 14:49:13 GMT

- reply

Satsman looks awesome! 🤩

2026-09-06 14:48:35 GMT

- reply

I guess it’ll depend on how the Bitcoin was acquired, if it was mined or bought tainted from an exchange and it’s got government tentacles all over it.
In many jurisdictions, spending Bitcoin is a taxable disposition.

2026-09-06 12:11:06 GMT

- reply

Presactly!

2026-09-06 02:24:06 GMT

They say “The government has a debt problem”
I hear “The government is financially irresponsible and also broke”

2026-09-06 02:18:16 GMT

- reply

Noice

2026-09-05 19:47:52 GMT

- reply

Having to switch contexts pushed me in this direction. I have a list of 20 items in my todo and I could do only 1 at a time. Of course I can use subagents which most harnesses already support.
I could have multiple installations/config folders I maintain manually but that just adds overhead and a perpetual config drift.
A subagent isn’t a full fledged agent, and if you want to resume its session later to do one more thing you can’t: you start from scratch again.
Then there’s the need for isolation: git worktrees aren’t true isolation, and harnesses for the most part reuse the same skills plugins etc and if you change the config folder you now have to manually authenticate everything for each new agent.

2026-09-05 16:25:52 GMT

- reply

LoRa ❤️
I should teach my agents how to use LoRa radios… hmm 🤔

2026-09-05 16:23:22 GMT

So now I’m moving from having a single agent that does everything to a primary I talk to that controls multiple ephemeral or static agents with similar capability as the primary. All comms happen over NATS so I don’t have to bother with “discovery” across hosts.

2026-09-05 16:21:13 GMT

- reply

I half expected someone to say Bitcoin is hot right now 😅

2026-09-05 16:20:35 GMT

- reply

I find these graph visualizations mesmerizing but not sure how useful it would be for me. I want the memory to be for the agent (which I’m sure it is). After a few minutes of watching the graph in its full glory I’m like: okay.

2026-09-05 16:18:55 GMT

- reply

I tried neo4j once and didn’t like it much. I’m using LatticeDB and it’s impressive how fast it is.
I paired with the Graphify skill and my agents are little Einsteins.
Best part: I don’t need no stinking flagship model from Anthropic to do this.

2026-09-05 16:17:37 GMT

- reply

I created a skill to have my agents meditate about previous sessions and propose improvements. It runs while I’m sleeping so I guess it’s a form of dreaming.