no-telemetry

name	no-telemetry
description	Disable and verify telemetry opt-outs for JS/TS libraries (Next.js, Prisma, Turborepo, Storybook, etc.). Use after scaffolding, after adding dependencies, or before opening a PR. Offline, no network, structured JSON + exit codes.

no-telemetry

Detect installed JS/TS libraries that collect telemetry and set the correct environment variables to disable it. Verify the result for CI and agents.

When to run

• After scaffolding a new app (create-next-app, npm create, etc.)
• After pnpm add / npm install / yarn add of framework or CLI tooling
• Before opening a PR, or as a CI step
• When the user asks to disable telemetry or prove it is off
• When you need to know which libraries are covered (list / why)

Commands (copy-paste)

# Apply opt-outs (non-interactive - always pass -y for agents)
npx no-telemetry init -y --json=compact

# Verify (exit 1 if anything still enabled)
npx no-telemetry check --json=compact

# Report only (always exit 0)
npx no-telemetry doctor --json=compact

# Discover coverage (no package.json required)
npx no-telemetry list --json=compact
npx no-telemetry why next --json=compact

Optional:

npx no-telemetry init -y --target .env.local
npx no-telemetry init -y --example          # commit-safe .env.example
npx no-telemetry init -y --target stdout    # KEY=VAL lines on stdout only
npx no-telemetry doctor --only installed
npx no-telemetry check --ignore netlify-cli
npx no-telemetry check --json --quiet       # compact + no stderr noise

Exit codes

Code	Meaning
0	Success (doctor always; check when policy passes; init applied or noop)
1	Policy failure - check found enabled telemetry; or user aborted init
2	Tool/usage error - bad flags, no package.json, or non-interactive init without --yes

Agents must pass --yes / -y for init. Non-TTY and CI=true sessions refuse to prompt.

Prefer --json=compact (or --json --quiet) for agent pipes - one line, no pretty-print padding.

Interpreting --json

Top-level shape (version: 1) for doctor/check/init:

• summary.enabled / summary.disabled / summary.applicable
• libraries[]: id, status, failsCheck, env[] with source (process-env | env-file | unset)
• actions[] on init (adds / placeholder fills / conflicts)

--only filters libraries[]; summary always describes the full post---ignore policy set.

list returns { version: 1, libraries: [...] } (full registry). why returns a single-entry object with id, env, docs, notes, …

Status tokens: disabled | enabled | not_applicable | not_found | unsupported.

check fails when any non-ignored library has failsCheck: true (status enabled).

Bindings may report multiple accepts values or "non-empty". Alternate signals use OR semantics by default. If a why / list row reports alternatePolicy: "fallback", its alsoSatisfiedBy values apply only while the primary env key is unset.

Rules

1. Never send network requests as part of this tool - it is offline by design.
2. Do not invent env vars - only write what init plans.
3. Do not overwrite conflicting values; report conflicts and let the human fix them.
4. Prefer committing opt-out vars that belong in the project (.env / .env.local / .env.example per team policy). Do not commit secrets; these opt-outs are not secrets.
5. After changing dependencies, re-run check --json=compact.
6. Use list / why <id> before guessing coverage.

Programmatic (Node)

import { scan, planInit, applyInit, failsCheck, buildReport, REGISTRY } from "no-telemetry";

const results = scan(process.cwd());
if (results.some(failsCheck)) {
  const plan = planInit(process.cwd());
  applyInit(process.cwd(), plan);
}