"Author, validate, troubleshoot, and adapt SkyPilot task and
Scanned 9/9/2026
Install to Claude Code
npx -y skills add VectorSpaceLab/AREX-Skill --skill task-yaml --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Task Yaml?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/vectorspacelab-task-yaml)More formats (shields.io, HTML) on the badges page.
---
name: task-yaml
description: "Author, validate, troubleshoot, and adapt SkyPilot task and
service YAMLs safely."
disable-model-invocation: true
metadata:
disco-role: operating
license: Apache 2.0
---
# SkyPilot Task YAML
Use this sub-skill when the user needs to write, review, convert, or debug a SkyPilot task YAML, a SkyServe YAML's task portion, or Python code that constructs `sky.Task` and `sky.Resources` from YAML-compatible config.
## Start Here
1. Read `references/yaml-reference.md` for accepted top-level fields, resource forms, service YAML fields, Python construction APIs, and safe validation flow.
2. Use `scripts/validate_task_or_service_yaml.py` for parser-only validation when SkyPilot is installed; it does not launch clusters, contact clouds, or write resources.
3. Read `references/troubleshooting.md` when validation fails or when YAML behavior is ambiguous.
## Owns
- Task fields: `name`, `workdir`, `num_nodes`, `resources`, `envs`, `secrets`, `api_server_access`, `volumes`, `file_mounts`, `setup`, `run`, and `config`.
- Resource fields: `infra`, `accelerators`, `cpus`, `memory`, `instance_type`, `use_spot`, `disk_size`, `ephemeral_storage`, `disk_tier`, `network_tier`, `ports`, `labels`, `any_of`, `ordered`, `autostop`, `job_recovery`, `priority`, and `priority_class`.
- Storage/file-mount basics: local path copies, cloud object-store mounts, `MOUNT`, `COPY`, `MOUNT_CACHED`, mount-cached presets, and Kubernetes volume references.
- Python parser construction with `sky.Task.from_yaml()`, `sky.Task.from_yaml_config()`, `sky.Task.from_yaml_str()`, and `sky.Resources(...)`.
- Service YAML field validation for the `service:` section and service-spec fragments, then routing runtime SkyServe operations elsewhere.
## Route Elsewhere
- Launching, executing, stopping, status, logs, autostop operations, and cleanup belong to `../cluster-operations/SKILL.md`.
- Managed job lifecycle, queues, recovery behavior, job pools, and job logs belong to `../managed-jobs/SKILL.md`.
- SkyServe operations, service updates, production rollout, and replica debugging belong to `../serving/SKILL.md`.
- Credential triage, provider setup, Kubernetes/Slurm/SSH details, GPU availability, and deep storage/provider failures belong to `../infrastructure-storage/SKILL.md`.
- SDK request handling, API server lifecycle, and remote server compatibility belong to `../sdk-api-server/SKILL.md`.
- Source edits, tests, formatting, and contribution rules belong to `../repo-development/SKILL.md`.
## Safe Validation
Run the bundled parser helper from this sub-skill directory or with an explicit path:
```bash
python scripts/validate_task_or_service_yaml.py --kind auto path/to/task.yaml
```
Use `--kind task` for normal task YAMLs and `--kind service` for SkyServe YAMLs or service-spec fragments. The helper imports SkyPilot parser APIs only; it is not a dry run, optimizer call, or credential check.
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!