Tencent Cloud TI-ONE Query Toolkit
Paste the following prompt into your AI chat to install this skill:
Please install @tencent-adm/tencentcloud-tione-skill according to https://skillhub.cn/install/skillhub.md.
About this skill
Problem
When operating Tencent Cloud TI-ONE, engineers often need to inspect training jobs, online services, notebooks, resource groups, model versions, datasets, logs, and events. Direct API calls or raw tccli usage can lead to inconsistent parameters, wrong ID scope, overly broad time ranges, or accidental write operations. This skill narrows common read-only queries to predefined bash scripts for troubleshooting and inspection.
How It Works
The skill runs scripts from scripts/ that call Tencent Cloud tccli using TENCENTCLOUD_SECRET_ID and TENCENTCLOUD_SECRET_KEY. It only executes Describe* APIs. It supports module queries such as describe-training-tasks.sh, describe-model-service-groups.sh, describe-notebooks.sh, describe-billing-resource-groups.sh, describe-datasets.sh, and describe-logs.sh. Use Status filters for state-dependent requests, AvailableNodeCount filters for resource groups, and instance-level IDs for logs and events. Fetch StartTime / EndTime from detail APIs to limit time ranges. generate-console-url.sh can generate console detail-page URLs.
Boundaries
The skill does not create, update, or delete cloud resources, and it does not allow bypassing scripts by calling SDKs or APIs directly. Out-of-scope requests should be reported as unsupported and directed to the Tencent Cloud console. The default region is ap-shanghai; supported regions include ap-beijing, ap-guangzhou, ap-shanghai-adc, ap-zhongwei, and ap-nanjing. If the user does not specify a region, warn before using the default. Credentials must not be printed or returned.
Use Cases
- Investigate failed training jobs by filtering status, then query instance-level logs and events.
- Inspect abnormal online services by checking service groups, service IDs, call info, and pod details.
- Count available GPU capacity by filtering resource groups with free nodes and reviewing node specs.
- Debug notebook issues by fetching runtime details, pod names, and time-scoped logs.
Best For
- MLOps engineers who need quick status checks for training jobs, inference services, and notebooks.
- Algorithm engineers debugging failures by filtering task status and pulling instance-level logs and events.
- Platform administrators auditing resource groups, datasets, model versions, and console entry points.
- SRE engineers inspecting service call info, pods, and time-scoped logs during incidents.
Related Skills
Detects AI-generated images from Stable Diffusion, Midjourney, and GPT-4o via Tencent Cloud image moderation and returns risk scores.
A log analysis tool combining deterministic rules and AI reasoning, with error severity levels, keyword filtering, real-time monitoring, webhook alerts, batch directory analysis, and path, SSRF, and XSS protections.
Comprehensive network engineering skill covering fault troubleshooting, technical consulting, and architecture design across routing and switching, wireless, security, cloud computing, and optical transport.
A Windows-only read-only storage scanner that finds large directories, grades risks, and opens an interactive local HTML report with one-click cleanup for safe green items.