An intelligent teaching assistant that automates homework grading, manages student rosters, and generates comprehensive class performance reports.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install tiantian-grader
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 tiantian-grader 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 tiantian-grader?
tiantian-grader is a robust automation tool designed to assist educators in the tedious process of grading assignments and tracking student performance. By utilizing the deepseek-reasoner model and high-accuracy recognition, this Openclaw Skills solution can process both individual pages and large batches of student work. It handles the entire lifecycle of an assignment, from initial roster synchronization to the final export of statistical data.
The skill provides a highly structured workflow that allows teachers to define assignment formats, verify recognized answers, and toggle between different grading philosophies. By maintaining a persistent local workspace, it ensures that student progress is tracked over time, building a valuable question bank and memory of classroom performance. This integration makes it a premier choice for those looking to enhance their productivity through Openclaw Skills.
tiantian-grader Use Cases
- High-volume grading of paper-based assignments converted to digital images.
- Generating automated statistical reports and CSV exports for class performance analysis.
- Maintaining a persistent digital memory of student rosters and historical grading data.
- Implementing different grading standards for various subjects, such as strict spelling checks or loose semantic matching.
How tiantian-grader Works
- The user triggers the initialization process within the Openclaw Skills environment to load student rosters and historical data.
- The teacher selects an assignment format, choosing between Type A (separate cover pages) or Type B (mixed pages) to guide the recognition engine.
- Standard answers are uploaded and processed using OCR, allowing the user to verify and adjust scores or grading modes.
- Student work is uploaded either sequentially for real-time feedback or in bulk for batch processing.
- The system compares student submissions against the verified answer key, applying either strict or loose grading logic as configured.
- A final summary is generated, and detailed results are exported to the local workspace in CSV and JSON formats.
tiantian-grader Setup
To deploy this skill, ensure your environment is configured with the necessary cloud credentials and model settings.
# Set required Tencent Cloud credentials for OCR features
export TENCENT_SECRET_ID="your_secret_id"
export TENCENT_SECRET_KEY="your_secret_key"
# Ensure the deepseek-reasoner model is accessible in your configuration
openclaw config set models.grading deepseek-reasoner
Note: The skill requires web.enabled to be set to true for file handling and session persistence within the Openclaw Skills framework.
tiantian-grader Data Schema & Taxonomy
The skill organizes data within a dedicated workspace (/root/.openclaw/workspace/teacher_tiantian) using the following structure:
| Directory | Purpose |
|---|---|
/homework/original |
Original uploaded images of assignments. |
/corrected/by_student |
Individual JSON files containing detailed grading results for every student. |
/questions/bank |
A persistent repository of recognized questions and answers for future use. |
/exports/csv |
Generated spreadsheets for easy integration with gradebook software. |
/memory/MEMORY.md |
A markdown-based log of all historical grading activities and milestones. |
tiantian-grader Advanced Features
- Multi-mode grading allowing users to switch between Strict (exact match) and Loose (flexible/semantic) evaluation per question.
- Compound question support for complex assignments featuring multiple sub-questions with independent scoring.
- Intelligent OCR caching and recognition confidence indicators to ensure high-quality data extraction.
- Automatic roster management that synchronizes student IDs with homework submissions for error-free reporting.
SKILL.md
Loading
Related Openclaw Skills
CesiumJS Development
An expert AI agent skill for building, debugging, and optimizing production-ready CesiumJS 3D geospatial applications.
OpenClaw Free Models Configuration
A comprehensive guide to integrating cost-free and trial-based LLMs like DeepSeek and GLM-4 into your Openclaw Skills ecosystem.
Feishu Task Control
A comprehensive interface for managing and terminating background subagent tasks directly within Feishu chat environments.
PDF Processing
A comprehensive suite of tools for extracting structured data from PDFs, merging documents, and automating form-based workflows.
News and Event-Driven Analysis
A sophisticated analytical engine that transforms raw financial news and corporate events into actionable investment intelligence.
OpenClaw Troubleshooting (CN)
A comprehensive technical guide and toolkit for diagnosing and resolving common errors within the OpenClaw AI agent environment.
Featured*
