Hacker Newsnew | past | comments | ask | show | jobs | submit | jonrosner's commentslogin

you could for example create a skill to access your database for testing purposes and pass in your tables specifications so that the agent can easily retrieve data for you on the fly.


I made a small mcp script for database with 3 tools:

- listTables

- getTableSchema

- executeQuery (blocks destructive queries like anything containing DROP, DELETE, etc..)

I wouldn't trust a textual instructions to prevent LLMs from dropping a table.


That's why I give the LLM a readonly connection


This is much better than MCP, which also stuffs every session's precious context with potentially irrelevant instructions.


They could just make mcps dynamically loaded in the same way no?


It is still worse as it consumes more context giving instructions for custom tooling whereas the LLM already understands how to connect to and query a read-only SQL service with standard tools


Oooooo, woah, I didn't really "get it" thanks for spelling it out a bit, just thought of some crazy cool experiments I can run if that is true.


it’s also for (typically) longer context you don’t always want the agent to have in its context. if you always want it in context, use rules (memories)

but if it’s something more involved or less frequently used (perhaps some debugging methodology, or designing new data schemas) skills are probably a good fit


one thing that I am missing from the specification is a way to inject specific variables into the skills. If I create let's say a postgres-skill, then I can either (1) provide the password on every skill execution or (2) hardcode the password into my script. To make this really useful there needs to be some kind of secret storage that the agent can read/write. This would also allow me as a programmer to sell the skills that I create more easily to customers.


I have no clue how you’re running your agents or what you’re building, but giving the raw password string to a the model seems dubious?

Otherwise, why not just keep the password in an .env file, and state “grab the password from the .env file” in your Postgres skill?


I am thinking of distributing skills that I build to my clients. As my clients are mostly non-technical users I need this process of distribution to be as easy as possible. Even adding a .env file would probably be too much for most of them. With skills I can now finally distribute my logic easily, just send the raw files and tell them to put it into a folder - done. But there is no easy way for them to "setup" the credentials in those skills yet. The best UX in my opinion would be for Codex (or Claude, doesn't matter) to ask for those setup-parameters once when first using the skill and process the inputs in a secure manner, i.e. some internal secret storage


That’s exactly what I do.


> there needs to be some kind of secret storage that the agent can read/write

Why not the filesystem?

I would create a local file (e.g. .env) in each project using postgres, then in my postgres skill, tell the agent to check that file for credentials.


after testing it for a little I am pretty disappointed. While I do get 90 token per second out of it from my M4 Pro which is more than enough for a real world use case, the quality is just not there. I gave it a codebase that it should analyze and answer me some questions and it started hallucinating right away. No replacement for a "real" coding agent - maybe for other agentic work like sorting emails though.


running it on my M4 @ 90tps, takes 18GB of RAM.


If it uses 18GB of RAM, you're not using the official model (released in BF16 and FP8), but a quantization of unknown quality.

If you write "M4", you mean M4 and not M4 Pro or M4 Max?


M2 Max @ 17tps btw


Self host Yaade. If you dont have a server run it locally in a Docker container.


Thats why I self host Yaade https://docs.yaade.io (shameless plug lol)


I can't reproduce this, even when forcing it not to reason:

"how many b's are in blueberry? don't think, answer fast."

blueberry • b → 1 (at position 1) • l → no • u → no • e → no • b → 2 (at position 5) • e → no • r → no • r → no • y → no

Total: 2 b’s.


Nice tool, congrats. Looks reasonable for the target audience. Is there a chord library for common songs? Or even sharing my chords with others?


A share button is pretty high on the priority list, as that's what musicians in bands would need the most.

As for libraries of charts, there are many other apps that do that very well already! I wanted to focus on fast editing of charts for musicians writing original songs, who are not really served that well by other apps at the moment.


Seems like NFC may be a nice way to share since musicians would most likely be physically close to each other. Congrats on shipping!


I haven't thought about that, thanks! I'll probably have to write my own NFC integration with Godot...


Wow thanks a lot! We played DD Poker during the Pandemic. It really saved us from going insane:)


That's great! You are the 2nd person who's told me that. You don't happen to know someone named Ryan in Oakland?


No, I am from germany. Also good job on programming a game that plays flawlessly on pretty much all systems after so much time. Also, online play via IP address was super easy.


Great update. Unfortunately, the limitation of push notifications to 10 members puts it into the B- category of self-hosted OSS for me. I understand that costs do exist by providing a push service, but forcing Orgs into the same plan as everyone else kind of feels like it's an intentionally set boundary to lurk people in and then force them to go all-in. I'd much rather see another tier that only includes the cost for push notifications.


E: just checked and they did update their pricing to be 3,50 now for the unlimited push. It's better, but still per-user-based. Which is still bad IMO. It says "Support Zulip's open-source development". I'd happily do that with a fixed price or an honest per-push price.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: