agentleFS
kotaroyamazakikotaroyamazaki/godot-agent-kit/skills/steam-tags/SKILL.md

steam-tags

Skill2 starsChanged 13 days agoMITView on GitHub

Choose, order and publish the Steam store tags for a game: read the game's own design docs, collect the tags of comparable games from their store pages, decide the 20 tags and their order against Steam's tag rules, then apply them in the Steamworks Tag Wizard with the browser and confirm they were saved. Use for "fix the Steam tags", "are our tags right", "which tags should we use", "reorder the tags", or before a Next Fest / launch. 「タグを見直して」「Steam のタグ大丈夫?」「タグ設定して」「タグの順番を直して」などで使う。

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Only the top 20 tags count.

    line 18
  2. Distinctive:

    Remove tags that are inaccurate.

    line 23
  3. Distinctive:

    Put it in the top 5.

    line 70

≈ “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

Steam tags

Tags decide which shelves, "More Like This" rows and recommendations a game appears in. Decide them from evidence (the game itself and what comparable games use), not from what sounds right.

Steam's rules (partner docs, store/tags)

  • Only the top 20 tags count. Fill all 20.
  • The top 5 describe the game and weigh the most. Put the most specific, high-information tags there (sub-genre, the hook), and generic ones (Indie, Singleplayer, Casual, Adventure) at the bottom.
  • For recommendations, overlap with similar games matters more than order, and rare tags weigh more than common ones. A rare tag that does not fit (e.g. Perma Death on a non-roguelike) pulls the game toward the wrong audience.
  • Remove tags that are inaccurate. Community tags also mix in after release; recheck 1–2 weeks after launch.
  • A demo inherits its base game's tags. Publishing tags is separate from the store page publish and its review, and takes effect without review.

Procedure

  1. Read the game. Its README, design doc and store copy (e.g. docs/DESIGN.md, docs/steam/store_*.md). Note the real mechanics: how failure works, length of a run, viewpoint, what is hidden from the player (do not spoil a hidden theme with a tag), and the AI-content disclosure.
  2. Read the current tags. Public page: run the script below with the game's appid. Not public yet: open the Tag Wizard (step 6) and read the summary on the right.
  3. Collect comparables. Find 6–10 similar games (https://store.steampowered.com/api/storesearch/?term=...&cc=US&l=english) and print their tags with the script. Count which tags most of them share.
  4. Decide 20 tags and their order, and write the reason for every addition and removal. Show the user the list before touching Steamworks.
  5. Check tag names. The wizard is in the account's language. English ↔ Japanese names (and tag ids): https://store.steampowered.com/tagdata/populartags/english and .../japanese. A tag missing from that list does not exist (e.g. "Spot the Difference").
  6. Apply in the Tag Wizard (https://partner.steamgames.com/taxonomy/tagwizard/<appid>, Claude in Chrome):
    • Steps: top-level genre (max 2) → genre → sub-genre → visuals → themes → features → players → other → user tags → order → publish.
    • Toggling a choice: click the input inside the visible div.choices div.choice whose text is the tag name (JS .click() is reliable; coordinate clicks miss when the page scrolls).
    • "Other" (step 8, search box to add any tag) is skipped by "Next"/"Back". Open it by clicking その他 / Other in the right-hand summary. Going back and forth can untick tags in that step: re-read the summary before publishing.
    • Ordering: #sortable is a jQuery UI list. Re-append its li elements in the wanted order, then jQuery('#sortable').sortable('refresh'); the summary on the right follows.
    • Press 公開 / Publish, read the confirmation dialog (it lists the final 20 and the community tags it will remove), and confirm only with the user's go-ahead.
  7. Verify. Reload the Tag Wizard and read the summary. The public store page catches up later (cache).
# Print a game's public tags in order (works for any appid; add &l=japanese for Japanese names)
python3 - 2653790 2903560 <<'EOF'
import re, json, sys, urllib.request
for a in sys.argv[1:]:
    req = urllib.request.Request(f"https://store.steampowered.com/app/{a}/?l=english",
        headers={"Cookie": "birthtime=0; lastagecheckage=1-0-1990; wants_mature_content=1"})
    s = urllib.request.urlopen(req, timeout=30).read().decode("utf-8", "ignore")
    m = re.search(r"InitAppTagModal\(\s*\d+,\s*(\[.*?\])", s, re.S)
    print(a, ", ".join(t["name"] for t in json.loads(m.group(1))) if m else "(no tags / not public)")
EOF

Lessons from past titles

  • Anomaly-spotting games (The Exit 8 likes) use Hidden Object (探し物) as their genre marker: 6 of the 10 main titles. Put it in the top 5.
  • Short = 短時間 in Japanese. Fits games that finish in about 10–30 minutes; short games travel well on streams.
  • Artificial Intelligence reads as "made with AI" next to an AI-content disclosure, and players punish that. Keep it only when AI is an open, central theme.
  • Controller adds almost nothing as a tag; Steam shows controller support separately.
  • Retro-console horror sellers (KinitoPET, Buddy Simulator 1984, FAITH) share Psychological Horror, Retro, Pixel Graphics and 1990's.

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.