send a workflow · the offer
Send us a workflow.
Describe one AI task you repeat, in ChatGPT, Claude, Cursor, Codex, or scripts. We convert the best ones into runnable .nika.yaml examples, credited to you.
what you send
Every Friday I paste the week’s merged PRs into a chat and ask for a changelog draft. It can read git log; it may only write CHANGELOG.draft.md.
plain words are enough
what ships back
# from @your-handle's friday ritual · converted + checkednika: v1workflow: id: friday-changelog description: "the week's merged PRs become a changelog draft"model: ollama/llama3.2permits: exec: ["git"] tools: ["nika:write"] fs: { write: ["./CHANGELOG.draft.md"] }tasks: collect: exec: command: ["git", "log", "--merges", "--since=7 days ago", "--oneline"] draft: with: merges: ${{ tasks.collect.output }} infer: max_tokens: 900 prompt: | Group these merges by area and write a changelog draft. ${{ with.merges }} save: with: draft: ${{ tasks.draft.output }} invoke: tool: "nika:write" args: { path: "./CHANGELOG.draft.md", content: "${{ with.draft }}" }clean: the engine’s checker, before a single token is spent
01 · you describe
Tell us the task
One AI task you repeat, in ChatGPT, Claude, Cursor, Codex, or scripts. Plain words are enough: what goes in, what should come out, what it may touch.
02 · we craft
We convert and check it clean
We write it as a runnable .nika.yaml: schema-true, audited before it runs (plan, cost, secrets), tested against the real engine.
03 · it ships
It lands in the gallery, credited to you
The best conversions ship as public examples, your name on the file. Your ritual becomes something anyone can run.
no form here · the funnel is public, on the repo · a GitHub account is the only thing we ask for