infra-skills
Open, portable agent skills for AI infrastructure engineering.
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Badge for your README
Markdown
[](/listings/open-infra-skills--infra-skills)HTML
<a href="/listings/open-infra-skills--infra-skills"><img src="/api/listings/open-infra-skills--infra-skills/badge.svg" alt="findsafeskills safety rating" /></a>Install
Install instructions from the README
As written by the repo's author. findsafeskills has not verified these steps.
The repository follows the open Agent Skills directory convention. Use the Skills CLI to inspect or install it:
# List all skills in the repository. npx skills add open-infra-skills/infra-skills --list # Install one skill interactively. npx skills add open-infra-skills/infra-skills --skill optimize-musa-training # Install one skill globally for Codex without prompts. npx skills add open-infra-skills/infra-skills \ --skill optimize-musa-training \ --global \ --agent codex \ --yes
You can also clone the repository and point a compatible agent at a skill directory directly. Each skill keeps its instructions and optional scripts, references, and agent metadata together.
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
README.md · README · 115 lines
View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)
- # Open Infra Skills
- Open, portable agent skills for building, operating, profiling, and optimizing AI infrastructure.
- This repository is intended to grow into a comprehensive, vendor-aware library for the full AI systems lifecycle: accelerator bring-up, distributed training, inference serving, kernels and compilers, data pipelines, observability, benchmarking, and cluster operations.
- ## Why This Repository Exists
- AI infrastructure work is full of environment-specific details, expensive experiments, and performance claims that are easy to mismeasure. Open Infra Skills packages repeatable engineering workflows as self-contained agent skills so that an assistant can:
- - establish a reproducible baseline before changing a system;
- - choose the right diagnostic layer, from application timing to system traces and kernel counters;
- - preserve numerical behavior while optimizing performance;
- - distinguish measured evidence from hypotheses;
- - work across local machines, SSH hosts, schedulers, containers, and clouds without assuming one client operating system.
- ## Skill Catalog
- | Skill | Category | What it helps with |
- | --- | --- | --- |
- | [`optimize-musa-training`](skills/accelerators/optimize-musa-training/SKILL.md) | Accelerators | Profile, debug, and optimize AI training on Moore Threads MUSA GPUs, including MFU/HFU, Moore Perf, FA2, `torch.compile`, FSDP, MCCL, and kernel-level analysis. |
- The machine-readable catalog is available in [`catalog.json`](catalog.json).
- ## Install
- The repository follows the open Agent Skills directory convention. Use the Skills CLI to inspect or install it:
- ```bash
- # List all skills in the repository.
- npx skills add open-infra-skills/infra-skills --list
- # Install one skill interactively.
- npx skills add open-infra-skills/infra-skills --skill optimize-musa-training
- # Install one skill globally for Codex without prompts.
- npx skills add open-infra-skills/infra-skills \
- --skill optimize-musa-training \
- --global \
- --agent codex \
- --yes
- ```
- You can also clone the repository and point a compatible agent at a skill directory directly. Each skill keeps its instructions and optional scripts, references, and agent metadata together.
- ## Repository Layout
- ```text
- infra-skills/
- |-- skills/
- | `-- <category>/
- | `-- <skill-name>/
- | |-- SKILL.md
- | |-- agents/ # Optional agent-facing metadata
- | |-- references/ # Optional progressive-disclosure docs
- | `-- scripts/ # Optional deterministic helpers
- |-- scripts/ # Repository validation and catalog tooling
- |-- .github/ # CI, issue forms, and PR guidance
- |-- catalog.json # Generated skill index
- `-- README.md
- ```
- The category layer keeps a large library navigable while remaining compatible with tools that discover `skills/<category>/<name>/SKILL.md`.
- ## Scope
- Planned categories include:
- - `accelerators`: vendor runtimes, device bring-up, profiling, and architecture-aware optimization;
- - `distributed-training`: DDP, FSDP, tensor/pipeline parallelism, collectives, and failure diagnosis;
- - `inference-serving`: throughput, latency, batching, caching, quantization, and serving engines;
- - `compilers-kernels`: graph compilers, custom operators, Triton/CUDA/MUSA kernels, and numerical validation;
- - `data-storage`: preprocessing, dataloaders, object storage, checkpoint I/O, and data movement;
- - `observability-benchmarking`: MFU/HFU, traces, rooflines, reproducible benchmarks, and cost analysis;
- - `cluster-operations`: schedulers, containers, networking, health checks, and incident workflows.
- This scope is a direction, not a requirement to create empty scaffolding. Categories are added when a maintained skill exists.
- ## Design Principles
- 1. **Evidence first.** Measure before optimizing, and state exactly what a metric includes.
- 2. **Correctness is a gate.** A speedup is retained only after numerical and behavioral checks pass.
- 3. **Portable orchestration.** Skills describe target-system behavior without requiring a particular laptop OS, IDE, shell transport, or scheduler.
- 4. **Progressive disclosure.** Keep the main `SKILL.md` focused and load deeper references only when needed.
- 5. **Reproducible experiments.** Record source revision, environment, command, workload shape, warmup policy, and raw evidence.
- 6. **Secure by default.** Never publish credentials, private hosts, proprietary data paths, or internal registry details.
- 7. **Honest negative results.** Document failed or neutral experiments when they prevent others from repeating expensive work.
- ## Quality Bar
- Every skill must:
- - have valid `name` and `description` frontmatter;
- - use a lowercase kebab-case directory matching its skill name;
- - contain actionable workflows rather than a loose collection of tips;
- - separate facts, measurements, inferences, and recommendations;
- - include verification steps and failure boundaries;
- - avoid private infrastructure details and embedded secrets;
- - pass repository validation and keep generated catalog data current.
- Run the checks locally with:
- ```bash
- npm test
- ```
- See [`CONTRIBUTING.md`](CONTRIBUTING.md) for the authoring and review workflow.
- ## Project Status
- Open Infra Skills is in early development. The first skill focuses on MUSA training optimization; the repository structure and validation pipeline are designed to support a much broader AI infrastructure collection without turning every skill into one giant prompt.
- ## License
- Licensed under the [Apache License 2.0](LICENSE).
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.