Skills Officially Comes to Codex • Hacker News Board

Re: Skills Officially Comes to Codex

#91

Post by didibus » Sat, Dec 20, 2025, 6:30 PM UTC

andybak wrote: post #20

Skills, plugins, apps, connectors, MCPs, agents - anyone else getting a bit lost?

None of them matter that much. They're all just ways to bring in context. Think of them as conveniences.

Tools are useful so the AI can execute commands, but beyond that it's just ways to help you build the context for your prompt. Either pulling in premade prompts that provides certain instructions or documentation, or providing more specialized tools for the model to use along with instructions on using those tools.

Re: Skills Officially Comes to Codex

#92

Post by NitpickLawyer » Sat, Dec 20, 2025, 6:32 PM UTC

fizx wrote: post #87

NitpickLawyer wrote: post #76

Earlier quoted context omitted.

On top of everything you've described, one more advantage is that you can use the agents themselves to edit / improve / add to the skills. One easy one to do is something like "take the key points from this session and add the learnings as a skill". It works both on good sessions with new paths/functionality and on "bad" sessions where you had to hand-hold the agent. And they're pretty good at summarising and extract…

Add reinforcement learning to figure out which skills are actually useful, and you're really cooking.

DSPy with GEPA should work nicely, yeah. Haven't tried yet but I'll add it to my list. I think a way to share within teams is also low-hanging fruit in this space (outside of just adding them to the repo). Something more org-generic.

Re: Skills Officially Comes to Codex

#93

Post by SatvikBeri » Sat, Dec 20, 2025, 6:33 PM UTC

hu3 wrote: post #81

jonrosner wrote: post #16

Earlier quoted context omitted.

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

Re: Skills Officially Comes to Codex

#94

Post by unbelievably » Sat, Dec 20, 2025, 6:37 PM UTC

zahlman wrote: post #88

colonCapitalDee wrote: post #79

Earlier quoted context omitted.

Before the first loop iteration, the harness sends a message to the LLM along the lines of. postgres Directions on how to query the pre-prod postgres db skills/postgres.md The harness then may periodically resend this notification so that the LLM doesn't "forget" that skills are available. Because the notification is only name + description + file, this is cheap r.e tokens. The harness's ability to tell the LLM "IMPO…

> The harness's ability to tell the LLM "IMPORTANT: this is a skill, so pay attention and use it when appropriate" and then periodically remind them of this is what differentiates ... And do we know how it does that? To my understanding there is still no out-of-band signaling.

[dead]

Re: Skills Officially Comes to Codex

#95

Post by shimman » Sat, Dec 20, 2025, 6:58 PM UTC

tacone wrote: post #89

I don't understand how skills are different than just instructing your model to read all the front-matters from a given folder on your filesystem and then decide if they need to read the file body.

Yes I'm confused as well, it feels like it's still all prompting which isn't new or different in the LLM space.

Re: Skills Officially Comes to Codex

#96

Post by afro88 » Sat, Dec 20, 2025, 7:00 PM UTC

zahlman wrote: post #88

colonCapitalDee wrote: post #79

Earlier quoted context omitted.

Before the first loop iteration, the harness sends a message to the LLM along the lines of. postgres Directions on how to query the pre-prod postgres db skills/postgres.md The harness then may periodically resend this notification so that the LLM doesn't "forget" that skills are available. Because the notification is only name + description + file, this is cheap r.e tokens. The harness's ability to tell the LLM "IMPO…

> The harness's ability to tell the LLM "IMPORTANT: this is a skill, so pay attention and use it when appropriate" and then periodically remind them of this is what differentiates ... And do we know how it does that? To my understanding there is still no out-of-band signaling.

A lot of tools these days put an extra message into the conversation periodically that the user never sees. It fights against context rot and keeps important things fresh.

Re: Skills Officially Comes to Codex

#97

Post by pests » Sat, Dec 20, 2025, 7:14 PM UTC

tacone wrote: post #89

I don't understand how skills are different than just instructing your model to read all the front-matters from a given folder on your filesystem and then decide if they need to read the file body.

That is basically what it is tho.

One difference is the model might have been trained/fine-tuned to be better at "read all the front-matters from a given folder on your filesystem and then decide..." compared a model with those instructions only in its context.

Also, does your method run scripts and code in any kind of sandbox or other containment or do you give it complete access to your system? #yolo

Re: Skills Officially Comes to Codex

#98

Post by user3939382 » Sat, Dec 20, 2025, 7:41 PM UTC

What they’re calling skills is a 5% weak implementation of what skills should be. My AI models fix this.

Re: Skills Officially Comes to Codex

#99

Post by mbreese » Sat, Dec 20, 2025, 7:44 PM UTC

shimman wrote: post #95

tacone wrote: post #89

I don't understand how skills are different than just instructing your model to read all the front-matters from a given folder on your filesystem and then decide if they need to read the file body.

Yes I'm confused as well, it feels like it's still all prompting which isn't new or different in the LLM space.

It’s all just loading data into the context/conversation. Sometimes as part of the chat response the LLM will request for the client do something - read a file, call a tool, etc. The results of which end up back in the context as well.

Re: Skills Officially Comes to Codex

#100

Post by mkagenius » Sat, Dec 20, 2025, 7:47 PM UTC

pupppet wrote: post #58

How are skills different than tool/function calling?

You can achieve what Skills achieve via function calling somewhat.

I've this mental map:

Frontmatter Name and arguments of the function

Text part of Skill md description field of the function

Code part of the Skill body of the function

But the function wouldn't look as organised as the .md, also, Skill can have multiple function definitions.