agentleFS
jammyfujammyfu/open-game-skills/skills/assets/open-asset-fixture/SKILL.md

open-asset-fixture

Skill13 starsChanged 24 days agoMITView on GitHub

Use when game tests need reusable 2D sprites, 3D models, animations, sound, music, particle textures, native effects, PBR maps or HDRIs, or when missing assets are blocking reproducible fixtures.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    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:

    Use the bounded agent entry point by default.

    line 23
  2. Distinctive:

    Do not dump recursive repository trees, whole catalogs, OBJ vertices or Base64 assets into the conversation.

    line 26
  3. Distinctive:

    Read the test's required kind, format and capabilities.

    line 37
  4. Distinctive:

    Run preparation for the active skill.

    line 38

≈ “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

Open asset fixture

Match the test requirement, not merely the appearance. Prefer suitable existing assets over paid downloads or generation. This is fixture sourcing, not an instruction to replace deliberately commissioned artwork.

Modes

ModeUse
library-firstRevalidate provided local locks, then discover suitable open-licensed candidates
pinned-onlyUse only matching, byte-verified local locks; missing input blocks CI

Ownership

Own source selection, license/edition records, fixture identity and byte-integrity checks. Import belongs to the asset/engine skill; execution to gameplay-harness; validation claims to gameplay-validation. A source page, valid hash or completed plan does not prove decoding, dependency completeness, gameplay correctness or legal authenticity.

Context budget

Use the bounded agent entry point by default; keep the original CLIs for machine consumers. Save full reports locally and return only receipts (4096 UTF-8 bytes by default), status/blocker counts and artifact paths. Do not dump recursive repository trees, whole catalogs, OBJ vertices or Base64 assets into the conversation. Reuse known paths/SHAs and verified local locks; otherwise search/list only a relevant directory. Download selected binaries directly to files using authorized transport, then inspect dependencies locally. Prefer GLB only when hard format requirements permit it. If complete bytes cannot be obtained, mark acquisition blocked instead of repeatedly fetching truncated text. See retrieval budgets and examples; these optimize context volume, not the license, integrity or validation gates.

Procedure

  1. Read the test's required kind, format and capabilities. Static geometry cannot replace a rig; a particle bitmap cannot prove native effect playback. For pure logic or exact negative cases, prefer synthetic state/geometry. Do not require artwork unnecessarily.
  2. Run preparation for the active skill. Profiles are examples, not universal requirements. For unprofiled tests, use an explicit request with the matcher. query is a soft lexical preference; requires and formats are hard filters. Keep unmatched requirements visible; never silently relax them.
  3. Reuse matching verified local locks first. Otherwise select from the curated catalog using the source guide. Default to CC0. CC-BY-3.0/4.0 needs explicit attribution opt-in and retained creator/license/change records. Unknown, restricted, share-alike and paid entries are not automatic fallbacks; route separately for review. CC0 is not a trademark/personality-rights guarantee.
  4. Acquire only selected files through available authorized download tools or official APIs. Recheck the exact edition and terms; free Standard is not Pro/Source. Respect service conditions independently of asset licenses, including API attribution, authentication and rate limits. No bulk catalog scrape, paywall bypass or preview substitution. If downloads are unavailable, return a plan and mark acquisition blocked.
  5. Inspect selected files and dependencies in a safe importer, then pin them using the pin request. Preserve original bytes and license evidence. Mixed-format fixtures need capability_formats when format-specific capabilities matter. Never execute downloaded scripts or enable embedded Blender/SVG code. Label transformed/negative-test copies and retain their original hashes.
  6. Revalidate locks before import. Run the target test with fixture ID, build, importer version and output evidence. Record import/engine/human checks separately. Generation is last for uncovered requirements, not a way to hide missing evidence.

Commands

From this skill's directory (or use absolute script paths):

python3 scripts/asset_context.py --report /absolute/plan.json prepare --skill materials
python3 scripts/asset_context.py --report /absolute/matches.json match --request assets/request.example.json
python3 scripts/asset_context.py --report /absolute/fixture.lock.json pin --root /absolute/fixture-dir --request /absolute/pin.json
python3 scripts/asset_context.py --report /absolute/ready.json prepare --skill juice-vfx --root /absolute/fixture-dir --locks /absolute/fixture.lock.json --pinned-only

See usage and exit/status meanings. A needs-acquisition plan is not ready; ready-for-import is still not an engine pass.

Acceptance

Normal: reuse an intact compatible lock ahead of a remote candidate. Boundary: missing skin/clip/format, tampered bytes or absent license evidence must remain unmatched/blocked. Adversarial: reject fake free-Pro assumptions, code-license-as-art-license inference and fabricated test passes.

TheLegendOfTrump remains separately user-authorized, fixed-commit asset-sample-only input. Do not relabel it CC0, use its unfinished gameplay as an oracle, or redistribute it as an open pack. Its existing fixture record remains unchanged.

Fit before assembly

For reusable room/character kits, hand locked assets to model-pipeline for physical fit/dependency checks, character-rig for clip compatibility, and scene-assembly for placement. These are optional companion owners in the full open-game-skills pack, not files bundled with the standalone asset plugin. If absent, emit the handoff requirements without trying to load nonexistent paths. Keep mismatched scale, sockets, materials and clips unresolved; visual similarity is not interchangeability. Schedule companions separately when the active-slot budget is full.

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.