agentleFS
xsollaxsolla/xsolla-ai-kit/skills/description-to-shop/SKILL.md

description-to-shop

Skill9 starsChanged 5 months agoMITView on GitHub

Turns a game description or store listing (App Store, Google Play, Steam) into the shop brief that shop-builder-assembly builds from. Reached through nocode-shop-setup; not an entry point. Owns intake and inference only: presets, confirmation, CLI writes and preview belong to shop-builder-assembly.

  • Reads credentials

The file

We split this skill into the rules it gives an agent

11 rules. About 0 are common, 5 are distinctive, and 6 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 6

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Before asking anything, read references/onboarding-intake.md and run this skill's scripts/read_onboarding.py --project-root …

    line 20
  2. Very specific:

    Use the publisher's working project, not the installed plugin directory.

    line 22
  3. Distinctive:

    Read references/intake-schema.md.

    line 35
  4. Distinctive:

    Never ask the publisher to repeat a usable recorded answer.

    line 37
  5. Very specific:

    Before asking anything, read the publisher's own store listing (App Store, Google Play, Steam), and only that: the recorded store_url or a …

    line 41
  6. Distinctive:

    When the publisher delegates routine choices ("decide for me"), choose the preset, layout, visual treatment and draft copy yourself and …

    line 47
  7. Very specific:

    When nocode-shop-setup started this run, its one batch already asked the intake: read those answers and ask nothing new, except about a …

    line 51
  8. Very specific:

    Follow the contract in shop-builder-assembly/references/shop-brief.md and write one version-1 JSON object:

    line 61
  9. Distinctive:

    Never include API keys, tokens, session cookies, passwords, or browser storage.

    line 91
  10. Very specific:

    Do not request a separate approval here and do not run legacy local assembly scripts.

    line 104
  11. Distinctive:

    If intake cannot produce a valid brief, report the missing facts.

    line 110

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Description to Shop

Convert a short game description into a valid shop-builder-assembly brief, then hand it over. Never call Shop Builder write commands here.

0. Reuse the confirmed onboarding intake

Before asking anything, read references/onboarding-intake.md and run this skill's scripts/read_onboarding.py --project-root <publisher-project-root>. Use the publisher's working project, not the installed plugin directory. The helper reads .xsolla/onboarding.json only. It does not read a build path from .env, and it makes no writes or API calls. On an error, stop and report it rather than restarting the interview. For portal, or when the user asked for a game home, or the existing site has /news, /rewards, and /store, hand off to game-web-portal and stop here. nocode-shop-setup asks store or game home when it is still open. A missing ledger permits normal intake from the conversation. A loaded ledger supplies already-confirmed answers. A value that contradicts the ledger without an explicit correction is a conflict: keep the ledger value, ask one question that shows both values, and wait for the answer.

1. Extract the description

Read references/intake-schema.md. Classify every required fact from the ledger and conversation as stated, safely inferred, missing, or conflicting. Never ask the publisher to repeat a usable recorded answer. Infer presentation choices such as the likely preset; never invent prices, catalog groups, dates, game facts, studio names, or translated copy.

Before asking anything, read the publisher's own store listing (App Store, Google Play, Steam), and only that: the recorded store_url or a link the publisher gives. Never search for it by the game's name. It supplies the name, platforms, developer, description, icon and screenshots; ask only for what it doesn't. Use its art and text in the brief: the plan shows them with the listing as their source, and the plan's yes is the only approval for them. When the publisher delegates routine choices ("decide for me"), choose the preset, layout, visual treatment and draft copy yourself and show them in assembly's plan instead of asking.

When nocode-shop-setup started this run, its one batch already asked the intake: read those answers and ask nothing new, except about a conflict. Otherwise, ask for missing required facts and unresolved conflicts in one batch. Use read-only discovery for the active CLI project and its catalog groups. A catalog that catalog-design has proposed and that waits for the same yes counts as existing. If the catalog is absent with no such proposal, or a named group is in neither, stop and point at the catalog skill. Never create entities here.

2. Produce the shared brief

Follow the contract in shop-builder-assembly/references/shop-brief.md and write one version-1 JSON object:

  • game — platforms and lifecycle.
  • site — proposed slug, locales, and auto or an explicit preset. On a site the developer made from a Publisher Account template, use that site's slug; any preset builds there.
  • catalog.groups — same-project groups that exist, or that catalog-design's pending proposal creates under the same yes; __all__ for all items of a supported type. Assembly's preflight checks every group exists before any site write.
  • catalog.liveops — the promo code or coupon promotions, reward chains and offer chains found by read-only xsolla liveops list-promo-codes, list-coupons, list-reward-chain and list-offer-chain. The lists include inactive entries: count only those with is_enabled: true. Also count the ones catalog-design's pending proposal creates and activates under the same yes, so their blocks are in the plan the developer approves.
  • content — publisher page overrides, and the copy the plan will show word for word: content.copy for block titles, subtitles and descriptions, content.buttons for the hero's button labels, and content.faq for the questions and answers, in the primary locale, as the shop brief reference describes. Fill content.seo by default, without asking: per page, a search title with the game's name, a one-line description from the listing, and the key art as the share image.
  • brand — the listing's art and app icon, unless the publisher supplied their own: download them next to the brief and list them in brand.images (key art on the leadGameSales background, the icon as the footer logo). Put the palette taken from that art in brand.theme, as the assembly skill's visual style describes.
  • sources — add {"kind":"description", ...}, and {"kind":"external_store", ...} with the listing URL when it supplied facts or art. Keep the provenance of any publisher answer used to fill a gap. Attribute reused intake to .xsolla/onboarding.json and its keys with kind: publisher_answers; retain provenance for corrections and inferences.
  • Never include API keys, tokens, session cookies, passwords, or browser storage.

Validate with the assembly skill's scripts/validate_shop_brief.py and fix input errors before handing off.

3. Delegate assembly

Invoke shop-builder-assembly with the validated brief. It exclusively owns preset selection and block defaults, the confirmation-bound plan, project and login preflight, backup-before-write, every xsolla shopbuilder write, structural, localization, catalog, readiness and preview verification, and the never-publish safeguard.

Do not request a separate approval here and do not run legacy local assembly scripts. The developer approves one plan, in shop-builder-assembly, shown in the same message as any pending catalog proposal.

Failure handling

If intake cannot produce a valid brief, report the missing facts. If assembly stops, return its exact blocker and backup location — do not retry or switch projects.

References

  • references/intake-schema.md — fields, inference boundaries, completeness gate.
  • references/plan-format.md — the handoff object and its rules.
  • shop-builder-assembly/references/shop-brief.md — authoritative contract.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.