Skip to main content

pai-configure-runtime

Configures Physical AI Runtime objects with physicalai.config and YAML recipes. Use when writing class_path/init_args for robots, cameras, runtime or inference components, exporting a local object with @export_config, or choosing jsonargparse for typed construction. DO NOT USE FOR editing src/physicalai/config or untrusted remote recipes.

소스 정보

저장소
openvinotoolkit/physicalai
최근 소스 활동
2026년 10월 15일 15:57
감지된 SKILL.md 언어
영어
스타
39
포크
30

설치 방법

기본적으로 소스를 먼저 확인하는 Prompt가 선택됩니다. 직접 명령으로 전환하거나 로컬 사본을 다운로드할 수도 있습니다.

소스 파일 검토

설치 여부를 결정하기 전에 SKILL.md와 SkillsMP에 표시된 보조 파일을 읽어 보세요.

SKILL.md 표시 중

SKILL.md
소스 지침 · 읽기 전용 미리보기
name
pai-configure-runtime
description
Configures Physical AI Runtime objects with physicalai.config and YAML recipes. Use when writing class_path/init_args for robots, cameras, runtime or inference components, exporting a local object with @export_config, or choosing jsonargparse for typed construction. DO NOT USE FOR editing src/physicalai/config or untrusted remote recipes.
license
Apache-2.0
# Working with `physicalai.config` Runtime owns `physicalai.config`; Studio consumes it as a package. Use trusted local configuration files only—`class_path` can import and instantiate Python objects. ## Workflow 1. **Pick the API** - Portable YAML recipes (robots, cameras, exported components): `Config` and `@export_config`. - Known Python types (trainers, dataclass configs, CLI models): jsonargparse (`ArgumentParser`, `add_class_arguments`, `parse_object`, `instantiate`). 2. **Author a recipe** — `class_path` + `init_args`; nest recipes only for trusted local config. See [instantiating components](https://github.com/openvinotoolkit/physicalai/blob/main/docs/how-to/config/instantiate-components.md). - Done when: dict/YAML passes validation without `ConfigError`. 3. **Export live objects** — `@export_config`, then `Config.from_instance(obj)` and `Config.save()`. Only trusted local sources. - Done when: the saved YAML reloads with `instantiate()` in a local smoke test. 4. **Typed construction** — use jsonargparse in your own CLI or app for typed arguments; avoid a generic untrusted YAML loader. 5. **Verify** — load the recipe in a short project-local smoke test with fake hardware, then use it with `physicalai run` only after reviewing every `class_path`. ## Validation loop Validate the local YAML against the [config schema](https://github.com/openvinotoolkit/physicalai/blob/main/docs/reference/config-schema.md) and round-trip a sample object in the user's Python environment. ## Required checks - Deeply nested config is rejected with `ConfigError` rather than silently truncated. - `class_path` comes only from trusted local config; never execute a downloaded recipe without review. - Studio and external plugins import `Config` from Runtime rather than copying the module. ## References - [Configuration explained](https://github.com/openvinotoolkit/physicalai/blob/main/docs/explanation/configuration.md) - [Instantiate components](https://github.com/openvinotoolkit/physicalai/blob/main/docs/how-to/config/instantiate-components.md) - [Config schema](https://github.com/openvinotoolkit/physicalai/blob/main/docs/reference/config-schema.md)
GitHub에서 보기