agentleFS
Sign inStart free
hyf020908hyf020908/modelscope-skills/skills/modelscope-jobs/SKILL.md

modelscope-jobs

Skill4 starsChanged 8 months agoApache-2.0View on GitHub

Execute plain-language batch-job requests as reproducible local or PAI DLC workflows and publish the outputs to ModelScope.

  • Reads credentials

The file

We split this skill into the rules it gives an agent

10 rules. About 0 are common, 5 are distinctive, and 5 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.

  • 5

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this skill when the user wants a batch workflow such as large-scale inference, data generation, statistics, or report production.

    line 8
  2. Very specific:

    If the run must go remote, prepare configs/remote.auto.env.

    line 30
  3. Very specific:

    If required PAI values are missing, create configs/pai.required.env, ask the user to fill it, and wait for continue.

    line 31
  4. Very specific:

    Create configs/pai.required.env with those four placeholders when needed.

    line 44
  5. Very specific:

    Put machine-readable results under data/ or another explicit output folder.

    line 48
  6. Very specific:

    Put human-readable summaries under reports/.

    line 49
  7. Distinctive:

    Never force the user to describe the job as a rigid prompt schema.

    line 61
  8. Distinctive:

    Never assume local paths exist remotely.

    line 62
  9. Distinctive:

    Never hide whether the job ran locally or on PAI DLC.

    line 63
  10. Distinctive:

    Never publish private raw data by default.

    line 64

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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

ModelScope Jobs

Use this skill when the user wants a batch workflow such as large-scale inference, data generation, statistics, or report production.

Request Style

  • Accept natural requests like:
    • Generate answers for this prompt dataset with Qwen2.5 and publish the result.
    • 统计这个 PDF 数据集的页面数并输出 markdown 报告。
  • Infer the lightest viable runtime:
    • local for small jobs
    • PAI DLC for long or heavy jobs

Primary Scripts

  • scripts/generate-responses.py
  • scripts/cot-self-instruct.py
  • scripts/finepdfs-stats.py
  • scripts/submit_pai_dlc.py

Workflow

  1. Translate the user request into one of the existing scripts or a short USER_COMMAND.
  2. Reuse local files when they already exist.
  3. If the run must go remote, prepare configs/remote.auto.env.
  4. If required PAI values are missing, create configs/pai.required.env, ask the user to fill it, and wait for continue.
  5. Upload local assets when the remote job needs them.
  6. Save outputs under stable local paths and publish only when credentials allow.

Credential Gate

For real PAI DLC submission, pause until these exist:

  • ALIBABA_CLOUD_ACCESS_KEY_ID
  • ALIBABA_CLOUD_ACCESS_KEY_SECRET
  • PAI_REGION
  • PAI_WORKSPACE_ID

Create configs/pai.required.env with those four placeholders when needed.

Output Rules

  • Put machine-readable results under data/ or another explicit output folder.
  • Put human-readable summaries under reports/.
  • Publish to ModelScope only when the target repo and token are available.

References

  • references/hardware_guide.md
  • references/hub_saving.md
  • references/troubleshooting.md
  • references/token_usage.md

Guardrails

  • Never force the user to describe the job as a rigid prompt schema.
  • Never assume local paths exist remotely.
  • Never hide whether the job ran locally or on PAI DLC.
  • Never publish private raw data by default.

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.

Check your setup, free