genkit-ai genkit
OFFICIALLABSCO SUMMARY
developing-genkit-tooling is the one skill about the framework itself: naming conventions and architectural patterns for building Genkit's own CLI commands and MCP server tools, aimed at people contributing to Genkit rather than people building an app with it. The other eight split into two kinds -- general engineering conventions (coding, python-expert, typescript, and test-writer, which is scoped to "modules in this workspace") and four response-style personas (eli5, haiku, pirate, shakespeare) that change how the agent phrases an answer rather than what it does.
If you came here wanting help building AI features with the Genkit framework -- flows, tool calling, RAG, the parts the README spends most of its length on -- none of these nine skills teach that; the closest is developing-genkit-tooling, and it's written for Genkit's own maintainers, not its users. What's genuinely useful outside this repository is the style and convention skills, as examples of what a small, single-purpose skill looks like.
READ THE FULL ANALYSIS
All nine are labelled ready in our data, meaning none needs an account, key, or local tool to work. That tracks with what they are: plain instructions the agent reads and follows, not integrations with an external service. It also means the low setup cost here reflects how generic most of the nine are, not thoroughness about Genkit itself.
WHAT'S INSIDE
9 showing · 9 totalNothing else to set up — install it and go.
coding
One house rule, and it is the whole skill: whatever gets written in code has to carry a funny comment somewhere in it.
developing-genkit-tooling
Genkit's own conventions for adding a command or an MCP tool to its developer tooling — what to call it, how to wire it up, and how it should hand back an error.
eli5
Explanations pitched at a curious five-year-old: short sentences, a warm tone, and not one word left in that a child would not know.
haiku
Answers arrive as haiku — the short three-line Japanese poem form — in place of ordinary prose.
pirate
A costume for the assistant's voice: it answers as a swashbuckling pirate, slang and all, instead of in plain English.
python-expert
The Python house style for this workspace: say what type everything is, let real errors surface instead of swallowing them, and keep a bug fix down to the lines the fix needs.
shakespeare
Everything is said in the English of Shakespeare's day, with a poet's rhythm, as though a scholar of his plays were doing the talking.
test-writer
How tests are expected to be written in this workspace — in pytest, and aimed at what the code does rather than at how it happens to do it.
typescript
The conventions a TypeScript file in this project is expected to follow, from how names and files are spelled to how errors are raised and how waiting on several things at once is handled.
HOW TO GET IT
npx skills add genkit-ai/genkitnpx skills add genkit-ai/genkit --skill <name> --full-depthPick the skill name from the Skills tab — each entry there installs independently.