Why oneliner101?

Most tutorials teach tools one at a time. The interesting work in 2026 happens when you compose them.

The meta-trick that started it

The day I piped claude --help into claude -p asking it to summarise its own flags, something clicked. Tools have grown faster than tutorials can keep up. Reading 70-line --help outputs is the old way. Piping them into an LLM that's seen a million similar docs is the new way:

$ claude --help | claude -p "
  Group these flags into 6 buckets (output, model, cost, tools, session, context).
  Markdown table per bucket. Mark the 3 flags I'll use daily with ⭐ and
  give me real-world examples with different option combinations.
"

That moment was the antidote to what I started calling the lazy-brain vibe-code syndrome — the urge to ask an AI agent to "just do it" without understanding what it's doing. The cure is one shell pipeline at a time. Read the data, read the schema, read the help text — then compose them. The LLM is the cofactor, not the driver.

oneliner101 is the catalogue of those compositions. Every entry is a real pipe you can run, copy, modify. None of them hide behind an agent. All of them teach a pattern.

The thesis

A single well-shaped pipeline — say, apify | jq | claude -p — can replace a 200-line Python notebook. It's faster to write, faster to read, and trivially shareable in a tweet.

oneliner101 is the catalogue: the 250 one-liners every data-driven builder should know in 2026. Hand-curated. No SEO-bait stubs. No "10 amazing Linux commands"-style listicles.

The promise: runs as-is

Every entry on this site works on a fresh laptop with brew install jq duckdb and a public-internet connection. Entries that need extra setup (Apify login, an LLM CLI key) are clearly tiered and labeled. There is never a hidden "my custom Postgres schema" dependency.

Curation roadmap

44 curated today — racing to 250 by July 31, 2026 🏁. The phases below are the climb, not the claim.

PhaseTargetTheme
v1 ← we are here44 curatedcurl + jq + LLM CLI + duckdb + apify + aws
v2 (June 2026)100+ ffmpeg, image pipelines, ML inference (Whisper, Ollama)
v3 (July 2026)200+ gcloud, kubectl, docker — cloud one-liners
v4 (July 31, 2026 🏁)250Community submissions, polish, full RSS + JSON feed

Support the project

oneliner101 is free and will stay free. If it saves you time, you can support the curation work:

Submit a one-liner

Have a pipeline that deserves a slot? The submit wizard drafts the JSON entry for you with Gemini 3.1 Pro, then opens a GitHub PR. The bot grades it (schema, shellcheck, jq syntax, safety, cost honesty, LLM rubric); the curator merges if it's clean. Criteria we hold submissions to:

  • Solves a real problem in one shell line.
  • Works on a fresh laptop (or clearly states what extra setup it needs).
  • Has a non-trivial pattern beyond the obvious — teaches something.
  • Cost per run is ≤ $0.10 or clearly labeled.
Open the submit wizard

built by @nnfuzzy · MIT licensed · last updated 2026-05-11