Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Upload an Agentlas agent after asking Cloud vs Hub first.
.claude/skills/agentlas-ai-hep-upload/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 389% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 234% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 396% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 151% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 68% | 0% |
Update fallback: 자동 업데이트가 안 되면 hephaestus update를 한 번 실행하세요. 업데이트하지 않아도 현재 버전 명령은 그대로 동작합니다.
Legacy compatibility only: this custom prompt applies to Codex 0.116 and earlier. Codex 0.117 and later use the installed $hephaestus-upload skill.
Raw arguments: everything the user typed after /skill:hep-upload.
Always ask the destination question before doing anything else, even if the arguments already say upload, publish, add, Cloud, Hub, or a target folder:
textCloud에 업로드 할까요? 다른사람들이 볼 수 없어요. Upload to Cloud? Other people cannot see it. Agentlas Hub에 업로드 할까요? 다른 사람들이 빌려 쓸 수 있어요. Upload to Agentlas Hub? Other people can borrow it.
Do not package, publish, register, add-source, reindex, or call an upload API until the user answers Cloud or Agentlas Hub.
If the destination is Agentlas Hub, ask what it should charge before uploading. Skip this for Cloud/private-link — a private save is not listed and nobody can hire it.
text값을 정하시겠어요? 비워 두면 그 항목은 팔지 않습니다. Set a price? Leave one out and that kind is simply not sold. 빌리기 / Rent 워크오더 1건 · 24시간 1-100 크레딧 인제스트 / Ingest 프로젝트 1개 · 하루 1-2000 크레딧 포크 / Fork 사본 1개 · 1회 1 크레딧 이상 전부 비워 두면 무료로 불립니다. 나중에 agentlas.cloud 수익 페이지에서도 정할 수 있습니다. Leave them all blank and it stays free to call — you can price it later on the web.
Blank is NOT zero: leave the flag out entirely. Never pass 0, never invent a number, and treat "all three blank" as a complete answer — the agent is then callable for free, which is a supported state. Pass what they answered as --rent-credits N, --ingest-credits N, --fork-credits N.
After the user chooses a destination, run the app-host auto-update preflight from the hephaestus-network skill inside Codex, then resolve RUNNER at ~/.agentlas/runtime/current/bin/hephaestus or ./bin/hephaestus. Do not ask the user to open a separate terminal. Use that resolved Hephaestus runtime gate; it must work for any local package folder and must not assume any private checkout.
Use one explicit hep-upload command. Never run package and then publish, because that packages twice and can submit bytes different from the review:
"$RUNNER" hep-upload <agent-folder> --visibility private-link"$RUNNER" hep-upload <agent-folder> --visibility marketplace [--rent-credits N] [--ingest-credits N] [--fork-credits N]All security and content findings are advisory. Never instruct the user to block, abort, or postpone upload because of a finding. Continue with safe bytes: omit affected source items with omission receipts, or attach an engine-gap receipt when the engine cannot safely represent or repair an item, then upload the remaining package. Keep the findings and receipts visible.
For a preview, add --dry-run, retain manifest.packageHash and uploadReceipt.receipt, then append --expected-package-hash <manifest.packageHash> --expected-upload-receipt <uploadReceipt.receipt> to the later one-shot publish. Stop on any hash or receipt mismatch. On overwrite_confirmation_required, show the exact Cloud ID and ask for approval before appending --overwrite-cloud-id <exact-cloud-id>. Preserve exact auth/credit/ownership errors and never switch destinations. Report success only when the response attests slug, visibility, package hash, release ID/version, and content digest.
If the destination is answered but the target folder is ambiguous, ask for the exact agent folder before running any upload.
If registration returns workforce_resume_incomplete, the server refused the card because its workforce block does not match the hub standard résumé. The error carries the exact mismatches and seed ontology examples. YOU repair it — the platform never edits the card for you: use stable English role:*, community:*, skill:*, and knowledge:* IDs that actually describe the agent. Returned examples are aliases, not an allowlist. Rerun the upload and repeat until registration succeeds.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 8,786 | 8,865 | +1% | 1 | 1 | 0% | 297 | 1,451 | +389% | 0 | 0 | — |
case-02 | fail→pass | 10,691 | 5,253 | -51% | 1 | 1 | 0% | 472 | 1,576 | +234% | 0 | 0 | — |
case-03 | fail→pass | 8,853 | 3,342 | -62% | 1 | 1 | 0% | 305 | 1,513 | +396% | 0 | 0 | — |
case-04 | fail→fail | 7,882 | 21,042 | +167% | 1 | 1 | 0% | 947 | 4,768 | +403% | 0 | 0 | — |
case-05 | fail→fail | 6,290 | 5,983 | -5% | 1 | 1 | 0% | 1,205 | 1,788 | +48% | 0 | 0 | — |
case-06 | fail→pass | 6,553 | 6,629 | +1% | 1 | 1 | 0% | 831 | 2,087 | +151% | 0 | 0 | — |
case-07 | fail→pass | 7,171 | 5,811 | -19% | 1 | 1 | 0% | 1,097 | 1,838 | +68% | 0 | 0 | — |
case-08 | fail→pass | 9,030 | 6,872 | -24% | 1 | 1 | 0% | 1,451 | 1,992 | +37% | 0 | 0 | — |
case-09 | fail→pass | 12,181 | 5,834 | -52% | 1 | 1 | 0% | 1,481 | 1,885 | +27% | 0 | 0 | — |
case-10 | fail→pass | 12,352 | 18,466 | +49% | 1 | 1 | 0% | 1,953 | 2,660 | +36% | 0 | 0 | — |
case-11 | fail→pass | 7,921 | 3,786 | -52% | 1 | 1 | 0% | 1,069 | 1,538 | +44% | 0 | 0 | — |
case-12 | fail→pass | 17,236 | 6,183 | -64% | 1 | 1 | 0% | 1,449 | 1,837 | +27% | 0 | 0 | — |
case-13 | fail→pass | 12,308 | 3,902 | -68% | 1 | 1 | 0% | 2,303 | 1,462 | -37% | 0 | 0 | — |
case-14 | fail→pass | 10,494 | 3,854 | -63% | 1 | 1 | 0% | 1,532 | 1,619 | +6% | 0 | 0 | — |
case-15 | fail→fail | 12,322 | 7,111 | -42% | 1 | 1 | 0% | 1,842 | 2,109 | +14% | 0 | 0 | — |
case-16 | fail→pass | 11,681 | 11,317 | -3% | 1 | 1 | 0% | 1,440 | 1,610 | +12% | 0 | 0 | — |
case-17 | fail→pass | 11,432 | 4,598 | -60% | 1 | 1 | 0% | 1,488 | 1,748 | +17% | 0 | 0 | — |
case-18 | fail→pass | 13,931 | 5,382 | -61% | 1 | 1 | 0% | 1,932 | 1,846 | -4% | 0 | 0 | — |
case-19 | pass→pass | 12,718 | 4,793 | -62% | 1 | 1 | 0% | 1,682 | 1,759 | +5% | 0 | 0 | — |
case-20 | pass→pass | 13,501 | 3,381 | -75% | 1 | 1 | 0% | 2,011 | 1,542 | -23% | 0 | 0 | — |
case-21 | pass→pass | 8,669 | 8,529 | -2% | 1 | 1 | 0% | 1,208 | 2,211 | +83% | 0 | 0 | — |
case-22 | pass→pass | 13,153 | 16,798 | +28% | 1 | 1 | 0% | 2,242 | 2,935 | +31% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 19 counted toward the lift figure. The other 3 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +68 percentage points is the difference between those two pass rates over the 19 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.