is-this-job-toxic 🚩

Paste a job posting. Get its red flags decoded into what they actually mean — with zero AI.

“We’re a family” → expect guilt trips for taking PTO and unpaid overtime. “Fast-paced” → understaffed and chronically on fire. “Competitive salary” → if it were competitive, they’d print the number. “Rockstar ninja” → we don’t actually know what this role is.

A tiny, 100% local, rule-based tool that scans a job description against a curated database of well-worn corporate red flags, scores its toxicity 0–100, and tells you — for each flag — what it really means. No API key, no model, nothing leaves your machine.

🌐 Try it in your browser →

Paste a posting, watch the flags light up. Nothing is uploaded — the whole engine runs client-side.

npx is-this-job-toxic "We're a family and we move fast. Looking for a rockstar who can wear many hats."

  100/100 ████████████████████ Run. 🏃🚩
  4 red flags in 16 words

  What they said → what it really means
    “We're a family” · “Family” boundaries
      ↳ Expect guilt trips for taking PTO and unpaid “for the team” overtime.
    “rockstar” · Vague role
      ↳ We don't actually know what this role is — but we want a hero for the price of one.
    “wear many hats” · Overwork
      ↳ We'll pay you for one job and have you do three.
    “move fast” · Always on
      ↳ Often ‘we move fast and break you.’

Why

Job postings have their own dialect. A lot of the friendly-sounding phrases are load-bearing euphemisms — they tell you about the pay, the hours, and the boundaries if you know how to read them. This decodes that dialect: it highlights the phrases, groups them by what they signal, and translates each one.

It’s a decoder, not a verdict machine. A great company can write a clumsy posting, and a toxic one can hire a good copywriter. Red flags are signals — use them to ask better questions in the interview, not to auto-reject.

Install

npm i -g is-this-job-toxic     # then:  is-this-job-toxic posting.txt
# or zero-install:
npx is-this-job-toxic posting.txt

Usage

is-this-job-toxic "paste the text right here"     # a string
is-this-job-toxic posting.txt                     # a file
pbpaste | is-this-job-toxic                        # the clipboard (macOS)
is-this-job-toxic posting.txt --md > report.md     # Markdown table
is-this-job-toxic posting.txt --json               # machine-readable

redflags is a shorter alias for the same command.

Flags

| Flag | What it does | | --- | --- | | --md [file] | Markdown table (What they said → what it really means) | | --json [file] | Full report as JSON | | --max-score <n> | Exit 1 if toxicity exceeds n — a CI gate for recruiters de-toxifying their own postings | | --ignore <category> | Skip a category (repeatable) | | --quiet | No pretty output (use with --max-score) | | --no-color | Disable ANSI colors |

For recruiters: lint your own postings

Run it on your job description in CI and fail the build if it’s too toxic:

is-this-job-toxic job-descriptions/backend.md --max-score 20 --quiet

The categories

| Category | Signals | | --- | --- | | Pay (hidden / low) | “competitive salary”, “commensurate with experience”, “equity-heavy” | | Overwork | “wear many hats”, “fast-paced”, “go above and beyond”, “self-starter” | | “Family” boundaries | “we’re a family”, “work hard, play hard”, “no egos” | | Vague role | “rockstar”, “ninja”, “guru”, “unicorn”, “jack of all trades” | | Always on | “evenings and weekends”, “around the clock”, “unlimited PTO” | | Passion-for-pay | “for exposure”, “must be passionate”, “rare opportunity” | | Desperation | “immediate start”, “urgent hire”, “must start ASAP” |

Library

The core is pure and browser-safe (no node:*), so you can use it anywhere:

import { analyze } from "is-this-job-toxic";

const report = analyze("We're a family and we move fast.");
report.verdict.score;   // 100
report.verdict.band;    // "run"
report.topFlags[0];     // { match: "We're a family", category: "culture", decode: "Expect guilt trips…", count: 1 }

How the score works

Each red flag carries a weight; toxicity is the accumulated weight, scaled and capped at 100. We deliberately don’t normalize by length — a red flag is a red flag whether the posting is 80 words or 800 — so the score stays predictable and we can always show you exactly which phrases produced it.

| Score | Band | | | --- | --- | --- | | 0–19 | Looks healthy | ✅ | | 20–44 | A few red flags | 🟡 | | 45–69 | Proceed with caution | 🚩 | | 70–100 | Run. | 🏃🚩 |

Privacy

Everything runs locally. The CLI reads your text, scores it in memory, and prints the result. The web playground runs the exact same engine compiled to the browser — open the network tab, you’ll see nothing leave. No telemetry, no account, no upload. Ever.

Contributing

The most useful contribution is a new red flag — a phrase, the category it signals, and an honest, funny-but-true decode. See CONTRIBUTING.md.

License

MIT © didrod205

A rule-based decoder, not a polygraph. It reads the clichés — your judgment reads the room.