A professional tool for AI agents to manage SJTU High-Performance Computing tasks including job submission, status monitoring, and data lifecycle management.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install sjtu-slurm-skill
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install sjtu-slurm-skill using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is SJTU SLURM Skill?
The SJTU SLURM Skill is a specialized integration designed for the Openclaw Skills framework, enabling AI agents to interact directly with the Shanghai Jiao Tong University (SJTU) High-Performance Computing platform, often referred to as 交我算. This skill acts as a bridge between high-level user requests and the technical complexities of SLURM-based clusters, allowing for seamless job querying, submission, and data management on behalf of the user.
By leveraging this skill within the Openclaw Skills ecosystem, developers can automate research workflows and manage large-scale computational tasks securely. It handles complex entry node selection, authentication via SSH certificates, and provides a structured way to interact with various node groups like pi, sy, and kp while maintaining high security standards for user credentials.
SJTU SLURM Skill Use Cases
- Automating the submission of SLURM jobs to specific partitions such as dgx2 or a100.
- Querying real-time job status and cluster resource availability.
- Managing data transfers between hot storage pools like Lustre or GPFS and cold archive storage.
- Retrieving storage quota usage and updating HPC account information via API.
- Requesting and refreshing SSH certificates for secure, passwordless cluster access.
How SJTU SLURM Skill Works
- The agent checks for a valid HPC API token in the workspace credentials directory.
- If needed, the agent requests a new token or SSH certificate, triggering 2FA via the JWB App or Email.
- The skill identifies the target node group (pi, sy, or kp) based on the user's requested partition or storage pool.
- It selects the appropriate entry node, such as pilogin or sylogin, for remote command execution.
- Commands are executed via SSH using the stored certificate and private key.
- The skill processes the output and provides the user with actionable feedback or data.
SJTU SLURM Skill Setup
To initialize this skill within your environment, ensure your workspace is configured correctly. Run the following commands to set up authentication:
# Step 1: Request an HPC API token
python req_token.py "<username>" --workspace "/path/to/workspace"
# Step 2: Request an SSH certificate (requires 2FA completion)
python req_certificate.py "<username>" --workspace "/path/to/workspace"
Ensure that the credentials directory is properly secured and that the Openclaw Skills agent has read/write permissions to the workspace.
SJTU SLURM Skill Data Schema & Taxonomy
The skill manages sensitive information and session data within a structured workspace. Key files and their roles include:
| File Name | Description |
|---|---|
credentials/hpc_token |
The bearer token used for calling the HPC REST API. |
credentials/private_key |
The user's private SSH key generated for remote access. |
credentials/certificate |
The signed SSH certificate allowing passwordless login to entry nodes. |
All remote paths follow the standard SJTU HPC storage convention, mapping user homes across /lustre, /dssg, and /archive.
SJTU SLURM Skill Advanced Features
- Intelligent entry node selection mapping partitions like a100 or huge to the correct node group.
- Support for virtual filesystems like /union to simplify data management across multiple cold storage pools.
- Robust error handling that directs users to specific documentation for identity binding or internal support.
- Automated token refresh logic to maintain long-running sessions within the Openclaw Skills framework.
- Tmux-wrapped data transfer operations to ensure reliability during large-scale file migrations on data nodes.
SKILL.md
Loading
Related Openclaw Skills
Bean-to-Soldier (Sa Dou Cheng Bing) Operation Distillation
A strategic operational framework designed to distill complex, large-scale project plans into reusable and standardized core skills for AI agents.
Instagram Bio Generator
An AI-powered professional bio creator designed to maximize social media discovery and follower conversion within the 150-character limit.
Browser Agent Pro
A professional-grade browser automation skill for AI agents featuring local headless execution and Browserbase cloud integration for anti-bot bypassing.
Feishu Chat Importer
A specialized utility to transform Feishu chat history exports into structured Openclaw episodic memory for AI-driven insights.
Education and Career Planning Assistant
A strategic AI career advisor that applies the methodologies of Zhang Xuefeng and Wu Liang to help students select college majors and plan professional lifecycles.
Hardware LLM Optimizer
A specialized tool that analyzes local hardware to recommend the optimal LLM size, quantization level, and deployment stack.
Featured*
