Rendered at 11:23:49 GMT+0000 (Coordinated Universal Time) with Cloudflare Workers.
angry_octet 1 days ago [-]
I can't think of anything worse than sharing skills via Dropbox. The version management and AIBoM problems that generates is extremely painful. There's no way to track which version LLM is being used or match it against the skill, and people will likely load up too many skills.
You don't have to expose git repos to end users to use git, or some other database, to provision skills.
dozerly 1 days ago [-]
It would probably be worse to fax them to your teammates
alwa 1 days ago [-]
I mean
Honestly from a BoM and version control standpoint, the fax might work out to be superior… they come timestamped; pop them in a filing cabinet, and…
angry_octet 21 hours ago [-]
Legitimately, it would work better. I could bulk fax Skills to each of the branch offices on Monday afternoon in preparation for the AI teleconference bridge meeting at 8AM. They can fax back any suggested improvements on Friday.
Because the thermal paper fades you would need freshly faxed Skills every few weeks or the turbo encabulator would misfeed when they were loaded. In fact, I'd have the local steno pool ladies feed in a weekly fax with authorisation codes for each Skill, and without that it just wouldn't load.
Y2K technology for the win.
detkin 1 days ago [-]
How do you concretely solve this for your marketing team? What’s the tech stack you have in place to support them?
nullpoint420 1 days ago [-]
Librechat self-hosted in our cluster that points to our ZDR ai gateway. They have SSO logins.
bberenberg 1 days ago [-]
Hey Dylan, long time. The solution we landed on was that the skills should be authored by a small set of people who know what they’re doing and made available to non technical team members “magically”. This means either use your IT systems to push the configs, or embed them into an agent that doesn’t live on the desktop. We’ve gone the latter route and are building a whole company around solving this for regulated customers.
detkin 1 days ago [-]
Hey man! Yeah, many of the non-techincal folks I've spoken to are using claude.ai or chatgpt.com. But I've been amazed at how many are using Claude CoWork or even Claude Code.
The theme seems to be wanting the same set of knowledge across any and every tool they use, without having to worry much about the mechanics of the how.
I agree that for security and governance conscious orgs a more robust server-side solution is probably what's needed. We've built that vault for sx as well. However, I am seeing that many larger orgs have decided to just build it themselves. There was a post from Mike at Gusto the other day saying as much.
The cost for build has just gotten so low now...
bberenberg 1 days ago [-]
Yup I agree with the consistent access across surfaces ideal. WorkOS has some cool piping to support this and uses it for their internal version of the same thing. I think everyone experimenting here is a good thing. I expect we’ll end up with a few good commonly use OSS approaches, a few tech companies with NIH syndrome versions for their needs, and commercial options like we have with past technologies. These are more reflections of culture amongst buyers than anything else.
detkin 1 days ago [-]
It also wouldn't surprise me if the model companies ship a strong version of this to try to lock us all in a bit more.
I hope that's not the case, or at the very least one storage and distribution system will work for all harnesses.
btown 1 days ago [-]
We've adopted a simple/similar Dropbox-based approach for skills and rules - each person's ~/.claude/skills is actually symlinked to a folder just for them inside a shared Dropbox folder, one that others on our (small) team can see and edit as well.
This solves a set of problems around people writing skills that reference artifacts or other skills that only exist on their system, and/or that reference their own name/information as the creator, and not knowing to make them self-contained and replicable. Luckily, adapting your colleagues' skills to self-contained versions and pulling them into your folder is trivial to instruct an agent to do. And you can have meta-skills that do this on the fly if a colleague has a skill that would unblock your project! (Editing to add a tip: make sure all the folders are set to offline visibility in Dropbox, rather than being loaded on demand from online.)
The courtesy simply has to be that you don't write into other people's skill folders unless/until they ask you to maintain something for them - at which point the words "I am assuming direct control" are said with all the necessary gravity and effect.
It's great to see someone putting UI and guardrails around this pattern!
detkin 1 days ago [-]
The one place I’ve seen this break down on teams is when a skill gets improved or adjusted. If everyone is copying it you end up not knowing which the best and most accurate. We worked with a team that had 5 copies that had all drifted.
That is why we built the deduplicator extension for sx. It finds the dups and lets you use the llm to build the consolidated “best” version of the skill.
btown 1 days ago [-]
To the point of people overtuning/overfitting skills that were once generic, to overly reference and draw upon their private preferences and workflows... sometimes having "installation = effectively forking" for a known-good skill is a feature, not a bug!
est 1 days ago [-]
We hosted shared skills via a git repo.
Simple pull & push would do.
detkin 1 days ago [-]
I agree that that works pretty well for developers who work with a code repository everyday. But, if you're working on a mono-repo, you can end up with more skills loading than you'd like pretty quickily.
Have you had success with non-technical people using git as their primary sharing source?
sdesol 1 days ago [-]
My extension for pi https://github.com/gitsense/pi-brains solves the too many skills problem and it can be adapted to work with any coding agent that supports hooks like Claude and Codex.
"read the file at data/accounting/q1.ledger and explain what this ledger tracks"
If you know what the use needs to read or edit, you can inject knowledge/skills for the agent.
trollbridge 1 days ago [-]
Git has easy to use GUI tools, particularly if you’re willing to use GitHub. I have not had trouble getting non technical staff to use it (book editors, graphic designers, writers, copywriters)
detkin 1 days ago [-]
Very cool, sounds like you have some technically open-minded co-workers.
Do you just create a claude or codex plugin in git for them? Since they likely aren't working against any code repos?
est 1 days ago [-]
yeah skills overwhelming is a problem. Splitting into sub-dirs works for now.
For us it's mostly developers.
detkin 1 days ago [-]
That makes sense. The one thing that really bugs me about git sharing is when you have different repos but want to share the same collection of skills. We have three different golang projects/repos. They follow very similar patterns and can share a bunch of skills but I don't want to copy/paste into the different repos and have them drift.
I also like having a system on top that manages our evals so I know when I can retire a skill that isn't pulling it's weight and I can see the usage stats to understand which skills are making a real difference.
try-working 1 days ago [-]
I feel that skills should live in the repo folder so they can be used by everyone on the project.
asp_hornet 1 days ago [-]
I landed on this too.
An advantage of this is when you got to hotfix an old customer, the skills align with how the code was authored at that point in time.
detkin 1 days ago [-]
A repo, if it’s not overly complex, is a great place to store skills for all the reasons you list. It’s just not the only place where skills are super valuable.
detkin 1 days ago [-]
I completely agree and sx tries to make it easier to share, not just on one project/repo but across teams.
For example, we have a collection of skills that have to do with brand marketing and blog writing. But developers use these too when blogging. That way all skills and knowledge in the org are shared. Not just with the folks who have git access and know where to look.
timmit 1 days ago [-]
Will a team wise GitHub repo a better solution?
- free version control
- one line set up
detkin 1 days ago [-]
sx has a git vault storage layer, or a local file layer or a server backed vault.
The thing it buys you over vanilla git is that you don't have to sym-link dirs for different AI harnesses. And, you can share skills across repos and teams without having to copy them into different repositories.
All that said, with the right setup, I think that vanilla git is a great answer. But if you start to want to bundle collections and share across teams and repos things start to fall apart.
Do you try to share across teams or repos? Or with non-technical teams?
prtmnth 1 days ago [-]
Why not a private GitHub repo? It's hard to believe that the technical team could not write 1 skill which all non-technical teams could use to sync their skills to GitHub.
dewey 1 days ago [-]
> And there was no way you were going to get your legal team to install and learn git.
8cvor6j844qw_d6 1 days ago [-]
Same thoughts. Get a team github repository and add the marketplace.
baalimago 1 days ago [-]
What's wrong with git?
MatekCopatek 1 days ago [-]
Absolutely nothing, if it fits your usecase, but how do you handle e.g. reusing the same skill in multiple repos?
Are you manually keeping all copies in sync?
dizhn 1 days ago [-]
Not that i recommend it or anything but git submodules would probably work.
InfiniteLoup 23 hours ago [-]
Who needs Dropbox? At my shop we're sharing skills with rsync...
You don't have to expose git repos to end users to use git, or some other database, to provision skills.
Honestly from a BoM and version control standpoint, the fax might work out to be superior… they come timestamped; pop them in a filing cabinet, and…
Because the thermal paper fades you would need freshly faxed Skills every few weeks or the turbo encabulator would misfeed when they were loaded. In fact, I'd have the local steno pool ladies feed in a weekly fax with authorisation codes for each Skill, and without that it just wouldn't load.
Y2K technology for the win.
The theme seems to be wanting the same set of knowledge across any and every tool they use, without having to worry much about the mechanics of the how.
I agree that for security and governance conscious orgs a more robust server-side solution is probably what's needed. We've built that vault for sx as well. However, I am seeing that many larger orgs have decided to just build it themselves. There was a post from Mike at Gusto the other day saying as much.
The cost for build has just gotten so low now...
I hope that's not the case, or at the very least one storage and distribution system will work for all harnesses.
This solves a set of problems around people writing skills that reference artifacts or other skills that only exist on their system, and/or that reference their own name/information as the creator, and not knowing to make them self-contained and replicable. Luckily, adapting your colleagues' skills to self-contained versions and pulling them into your folder is trivial to instruct an agent to do. And you can have meta-skills that do this on the fly if a colleague has a skill that would unblock your project! (Editing to add a tip: make sure all the folders are set to offline visibility in Dropbox, rather than being loaded on demand from online.)
The courtesy simply has to be that you don't write into other people's skill folders unless/until they ask you to maintain something for them - at which point the words "I am assuming direct control" are said with all the necessary gravity and effect.
It's great to see someone putting UI and guardrails around this pattern!
That is why we built the deduplicator extension for sx. It finds the dups and lets you use the llm to build the consolidated “best” version of the skill.
Simple pull & push would do.
Have you had success with non-technical people using git as their primary sharing source?
You can find a simple example at https://github.com/gitsense/gsc-rules-demos which shows how skills can be injected when needed. The example is:
"read the file at data/accounting/q1.ledger and explain what this ledger tracks"
If you know what the use needs to read or edit, you can inject knowledge/skills for the agent.
Do you just create a claude or codex plugin in git for them? Since they likely aren't working against any code repos?
For us it's mostly developers.
I also like having a system on top that manages our evals so I know when I can retire a skill that isn't pulling it's weight and I can see the usage stats to understand which skills are making a real difference.
An advantage of this is when you got to hotfix an old customer, the skills align with how the code was authored at that point in time.
For example, we have a collection of skills that have to do with brand marketing and blog writing. But developers use these too when blogging. That way all skills and knowledge in the org are shared. Not just with the folks who have git access and know where to look.
- free version control - one line set up
The thing it buys you over vanilla git is that you don't have to sym-link dirs for different AI harnesses. And, you can share skills across repos and teams without having to copy them into different repositories.
All that said, with the right setup, I think that vanilla git is a great answer. But if you start to want to bundle collections and share across teams and repos things start to fall apart.
Do you try to share across teams or repos? Or with non-technical teams?
Are you manually keeping all copies in sync?