Skip to content

t1k:marketing:store:auto-gen-policy-privacy

FieldValue
Modulestore
Version0.14.9
Effortlow
Tools—
/t1k:marketing:store:auto-gen-policy-privacy

Generates a long (~1,600–1,800 words), store-ready, company-wide privacy policy for a mobile-game publisher / developer store account — the policy you host and point Play Console and App Store Connect at.

Every run produces different wording, so no two published policies are identical: phrasing variants inside every paragraph, bullet order, middle-section order, optional sections, heading label style (1. / A. / none), the title, and the header/date format are all randomised (scripts/templates.cjs holds the variants, scripts/gen.cjs picks them).

The seed is printed to stderr; --seed <v> reproduces one policy exactly. Run twice with the same --seed to get matching HTML and MD from one random draw.

Terminal window
bun <skill>/scripts/gen.cjs \
--dev "Acme Games Ltd" --email support@acmegames.com \
--sdks admob,applovin,unity,mintegral,ironsource,meta,firebase,adjust,googleplay \
[--app "Block Puzzle"] [--date 2026-10-21] [--format html|md] [--out policy.html]

--format defaults to html (the hosted artifact); print to stdout when --out is omitted. --date defaults to today and must be YYYY-MM-DD. Exit 2 on an unknown --sdks key, a missing --dev/--email, or a malformed date — the error names the known keys.

Known --sdks keys (the SDKS map in scripts/gen.cjs, each rendered as a named link plus its privacy URL):

googleplay · admob · firebase · applovin · unity · mintegral · ironsource · meta · dtexchange · pangle · liftoff · inmobi · adjust · appsflyer · gameanalytics · cloudx

  1. Ask the user (AskUserQuestion) for the exact registered company name, the contact email, and which ad/analytics SDKs the games use. Use the company name verbatim — if it already ends in a legal suffix, flag a duplicated form such as “Ltd Limited” back to the user rather than publishing it.
  2. They do not know the SDK list → default to the typical MAX mediation set: admob,applovin,unity,mintegral,ironsource,meta,firebase,adjust,googleplay. Say that this is the default and that a missing network under-discloses.
  3. Generate the HTML (for hosting) and the MD, one run each with the same --seed, and open the HTML for review.
  4. Report the seed, so the exact policy can be regenerated later.
  • Varying wording never varies meaning. The template variants are same-meaning rewrites; every policy asserts the same facts. To add variety, add more variants to scripts/templates.cjs — do not reword a claim into a different claim.
  • The text asserts: no precise GPS location, no account required, an EU/UK consent prompt for personalised ads, and no targeting of children under 13. If a game differs, edit the output before publishing — the generator does not read the game’s actual data practices.
  • Not legal advice. A generated policy is a starting draft; the studio’s own review stands between it and a live listing.
  • Adjust’s privacy URL returned HTTP 429 during verification — rate-limited, not a confirmed URL. ByteBrew is omitted: its policy URL 404s. Re-check both before adding them.
  • Never copy an existing policy verbatim. The templates were written fresh after studying the studio’s example policies in the Drive folder; they are a reference for shape and coverage, never a source to paste.

Generation is deterministic script work, but the inputs are the user’s answers and the delivered policy is published, so the ask-and-review steps stay on a reasoning tier (routing SSOT: rules/orchestration-rules.md § “Task-Type → Agent Routing”).

StepAgent
Run scripts/gen.cjs with values the user already supplied and report the seed, the output paths and the word countt1k-fact-fetcher — local file generation, no network, nothing published

Stays reasoning-tier: collecting the company name / email / SDK list, and editing the output before it is published.

scripts/gen.cjs (CLI, seeding, HTML/MD renderers), scripts/templates.cjs (every wording variant — the only file to edit when adding variety).