shouldi 🔮

Should I…? — a sarcastic, non-blocking DevX oracle for the moments every developer knows better: deploying on Friday, merging that "quick fix", rebasing a fermenting branch, trusting AI-generated code. Giant weird verdict. Tiny punchline. Exit code 0.

npx @shouldi/cli deploy      # or: npm i -g @shouldi/cli && shouldi deploy

THE BUTTON LOOKS SMUG.
Do not reward it.

Verdict: emotionally no
Exit code: 0

Status: 🌱 Planning. A working website prototype exists (~270 original quotes, US-holiday-aware risk theming, screenshot export) — see docs/research/should-i-deploy-oracle.html. The npm @shouldi org scope is secured; @shouldi/cli is the umbrella package (unscoped shouldi is blocked by npm's punctuation-stripping similarity rule — see the validation notes). No functional code here yet — see the planning notes. First oracle to ship: shouldi deploy.

The family

One CLI, many oracles — each a "Should I…?" question for a risky developer moment:

| Command | Also as (planned aliases) | The moment | |---|---|---| | shouldi deploy | @shouldi/deploy | The flagship. Friday. Holiday eve. You know. | | shouldi merge | should-i-merge | That PR with "quick fix" in the title | | shouldi rebase | should-i-rebase | The branch has been fermenting for weeks | | shouldi ai | should-i-trust-this | The diff an AI wrote with total confidence | | …and more | | migrations, deps, infra, push, on-call — see the family PRD |

It does not block by default. It judges.

Where to look

All current thinking lives in docs/research/ — the origin PRD, the working website prototype, the validation/naming session, and the full oracle-family PRD. Start with the research index.

Conventions

Personal project under sharvilk — not affiliated with any employer. Follows the personal-repo conventions in ~/sharvil/README.md: own git repo, personal git identity, and — once implementation starts — the full engineering-standards baseline (TypeScript strict, lint, format, tests, CI). License: Apache-2.0.