Omniscient

Inspiration

Skills are a great standard for in context learning (pull instructions into context whenever needed). But they are random markdown files in github repos. We gave them a home (your graph), a public registry, and a way to learn new things and connect with other skills instantly.

What it does

Registry of skills
Skills are linked -> represented in database as links -> creates a graph
Agents can easily iterate through the registry through semantic search and more, and can traverse the graph to optimize it
Agents themselves can edit and create new skills (for example directly from claude code)

Customization is fundamental: skills give great standards, and can be simply imported, but they can also be customized on your own needs by talking with an ai agent in the platform, OR from your own ai agent (like claude code)

How we built it

Nextjs
Honojs
A cli with authentication that syncs to all your agents
Trpc
Better Auth
Modif

Accomplishments that we're proud of

Skills are all represented as a graph: we have the biggest STRUCTURED database of skills (not just a list of files)
It's so easy to create skills now that we created skills for almost every single hackeurope sponsors (ask us if interested)

What we learned

turning file links into a graph represented in a database is tough, but we were able to do it

Roadmap

create the biggest repository of skills
make it the world standard for skills management
security audit for public repository
we couldn't finish the stripe integration, but this is the business plan: Freemium model:
free plan: up to 10 skills (and unlimited sub-skills) and 100 ai messages
pro plan: unlimited skills and up to 1k ai messages Enterprise plan: custom b2b skill creation for big companies and environment and permissions setup