Skip to content
Back to skills

Serving Tools

ASecurity

Package, publish, inspect, and troubleshoot MMDetection3D serving utilities.

  • 247 stars
  • 0 votes
  • 0 copies
  • 2 views
  • Added September 8, 2026
devopsdockertestingapi

Works with

  • api

Security analysis

A100/100

Pro scans all 5 files and shows the line behind each finding

Scanned September 8, 2026

npx -y skills add VectorSpaceLab/AREX-Skill --skill serving-tools --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Serving Tools?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Serving Tools
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/vectorspacelab-serving-tools/badge)](https://www.skillsdirectory.com/skills/vectorspacelab-serving-tools)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: serving-tools
description: "Package, publish, inspect, and troubleshoot MMDetection3D serving utilities."
disable-model-invocation: true
metadata:
  disco-role: operating
license: Apache 2.0
---

# MMDetection3D serving-tools sub-skill

Use this sub-skill when the task is one of:

- package a checkpoint for TorchServe-style serving,
- verify the packaging inputs before export,
- publish or convert a checkpoint for release,
- inspect training logs, FLOPs, or throughput,
- print configs or fuse Conv+BN for utility work.

This sub-skill is safe by default. The bundled checker only validates paths and required artifacts; it does not start servers, build containers, download weights, or run model execution.

## Start here

1. Identify the utility family: serving packaging, checkpoint publishing/conversion, log analysis, FLOPs/throughput, or config inspection.
2. Before packaging, run [`scripts/check_serving_artifacts.py`](scripts/check_serving_artifacts.py) to confirm the config, checkpoint, handler, model name, and export target are ready.
3. For the serving flow and handler assumptions, read [`references/serving.md`](references/serving.md).
4. For the utility script catalog and safe-use boundaries, read [`references/tools-reference.md`](references/tools-reference.md).
5. For common failures and stop conditions, read [`references/troubleshooting.md`](references/troubleshooting.md).

## Route away

- Choosing a model family, config alias, or checkpoint family belongs in `configuration-model-zoo`.
- Raw inference demo commands or inferencer APIs belong in `inference`.
- Training, testing, evaluation, TTA, distributed launch, and Slurm workflows belong in `training-evaluation`.
- Geometry, box conversion, coordinate systems, and visualization internals belong in `structures-visualization`.
- Dataset preparation and conversion belong in `data-preparation`.
- Custom component implementation belongs in `customization-extensions`.

## Minimal operating rules

- Do not run TorchServe, Docker, download checkpoints, or benchmark jobs unless the user explicitly asks for execution.
- Treat the bundled checker as a preflight only; it prints missing artifacts and exits, but it does not repair anything.
- Keep utility advice tied to local files and explicit config/checkpoint pairs.
- Prefer the bundled references over free-form guesses when the user asks about script flags or expected artifacts.

Files in this skill

  • SKILL.md2.4 KB
  • references/serving.md2.6 KB
  • references/tools-reference.md3 KB
  • references/troubleshooting.md2.5 KB
  • scripts/check_serving_artifacts.py4 KB

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.