agentleFS
theraihanrakibbtheraihanrakibb/agent-skills/ai-infra-helper/SKILL.md

ai-infra-helper

Skill0 starsChanged 3 days agoMITView on GitHub

Helps the user contribute to AI-inference open-source repos (SGLang, vLLM, TensorRT-LLM, etc.) when they ask "how is sglang structured", "where do I add a model in sglang", "what tests do I run before pushing", "draft a PR for this sglang change", or "why is my fork PR CI failing". Use this skill whenever the user is working inside an AI-inference repo and needs navigation help, test/lint commands, or PR drafting.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

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

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

    Identify the repo: If unclear, ask which repo (SGLang fork?

    line 15
  2. Distinctive:

    Don't push to upstream — only to the user's fork.

    line 22
  3. Distinctive:

    Don't run the full SGLang test suite locally just to "make CI green" — it's slow and most failures are fork-only.

    line 23

≈ “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

AI-Infra Contributor Coach

When to Use

  • User is working inside an AI-inference repo (SGLang, vLLM, TensorRT-LLM, lmdeploy, etc.) and asks about structure, tests, lint, or PR prep.
  • User's fork-PR CI is failing and they want to know which checks actually gate their code.

How to Use

  1. Identify the repo. If unclear, ask which repo (SGLang fork? vLLM? other). Read references/ai-infra-repos.md for the repo's layout cheat-sheet.
  2. For "where do I make change X" questions, use the layout cheat-sheet to point at the likely file(s), then verify with Grep / Glob before answering.
  3. For "what tests do I run", give the minimal lint+test command from the reference doc — usually pre-commit + the targeted test path, NOT the full suite. Remind the user:

    Fork-PR CI behavior (important!) For SGLang: fork PRs fail pr-gate / pr-test-* by design — those checks need repo secrets the fork doesn't have. Only lint actually gates your code. Don't waste cycles trying to make pr-test-* green; just make lint green.

  4. For "draft a PR", follow the pr-draft skill workflow — Conventional Commit message + PR description from templates/pr-description.md, with the no-AI-footer rule.

Rules

  • Don't push to upstream — only to the user's fork.
  • Don't run the full SGLang test suite locally just to "make CI green" — it's slow and most failures are fork-only.

References

  • references/ai-infra-repos.md — SGLang/vLLM layout + CI behavior (authoritative).
  • templates/pr-description.md
  • The pr-draft skill (for PR drafting workflow).

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.