About SkillCannon

SkillCannon is a curated collection of AI skills for SaaS and product teams. The market already has scraped mega-catalogs, one of them with over a million entries. We are not trying to be the biggest list. We are trying to be the one you can trust.

What makes this different

Two things. First, curation with a point of view: every skill carries a VeryCreatives' take, an honest verdict from people who have actually used it, not an auto-generated blurb. Second, cross-agent compatibility: we check each skill against Claude Code, Codex, and OpenCode and label what works, what works with caveats, and what does not, with the caveats spelled out. A scraped catalog cannot cheaply copy either one.

We organize skills by the role that needs them, product, engineering, design, marketing, and operations, so you can find what is useful for your work instead of scrolling a wall of entries.

How we curate

Skills are discovered, enriched, and then reviewed by a human before anything publishes. Nothing reaches this site on autopilot. Compatibility flags start AI-inferred and carry an honest not-yet-verified note until a person confirms them, because "the model thinks it works" and "we ran it" are different claims and we will not blur them.

Every skill we publish gets a written VeryCreatives' take. Where we have run a skill in production, we add a "How VeryCreatives uses it" section with the real workflows and the honest caveats. Where we have not, we say so. The take is the part a scraped list cannot fake, so it is the part we will not skip.

Who is behind it

SkillCannon is built and maintained by the team at VeryCreatives, a product studio that has shipped SaaS for a living since 2013, serving clients across the EU, the US, and the UAE. Over that time we have been the tech and product team behind multiple successful startup exits. It is curated by Máté Várkonyi and Ferenc Fekete, co-founders of VeryCreatives, with takes contributed by the wider VeryCreatives team.

We do not just write about agent skills, we practice agentic coding as our craft. Proper rules, company-wide ADRs (architecture decision records), and senior and architect-level supervision are how we get very high quality code out consistently instead of occasionally. We are also big fans of OpenCode, which is part of why cross-agent compatibility is a first-class concern here and not an afterthought.

We put this to work for operationally complex businesses, not only startups, and often in regulated environments where the quality bar and the cost of a mistake are both high. The recommendations here are the tools we reach for in that work. The skills we call top picks earned it by saving us real time, and the takes come from using the tools, not skimming their READMEs. We also maintain an open awesome list of agent skills for SaaS on GitHub.

Start here

New to agent skills? Read what are Claude skills. Ready to browse? Start with the curated marketplace.