A shell-native spreadsheet CLI designed for AI agents to perform precise workbook inspection, manipulation, and verification.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install agent-sheet
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 agent-sheet 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 agent-sheet?
agent-sheet is a robust, local spreadsheet CLI specifically engineered for agentic workflows within the Openclaw Skills ecosystem. Unlike basic CSV parsers, it provides deep access to real workbook structures, including formulas, sheet lifecycles, and precise range management. It allows AI agents to interact with Excel and spreadsheet files directly from the command line, ensuring high-fidelity data processing without the overhead of external libraries.
By providing a safe shell roundtrip for workbook data, agent-sheet enables developers to stream data through standard Unix tools like awk or sed and write verified results back into the original workbook. It is the preferred tool when tasks require workbook-native scripting, formula fills, or structural verification that metadata-only tools cannot provide.
agent-sheet Use Cases
- Resolving workbook structures and maintaining target consistency using unique entry IDs.
- Inspecting complex spreadsheet elements including hidden sheets, ranges, formulas, and search hits.
- Writing sparse cells, bounded ranges, or anchored review tables into existing workbooks.
- Managing full file lifecycles, including importing local workbooks for editing and exporting final versions.
- Building automated verification passes to ensure data integrity after complex shell transformations.
How agent-sheet Works
- Preflight and Resolution: The agent resolves the workbook context and establishes a stable entry ID to ensure all subsequent commands target the correct file.
- Inspection: Using commands like sheet list or inspect workbook, the agent analyzes the existing structure and formula state.
- Command Selection: The agent selects the most granular command (e.g., write cells vs. write range) to minimize side effects on the workbook.
- Execution: The CLI executes the mutation, potentially piping data through shell tools for advanced processing.
- Verification: A final verification pass is performed using specialized scripts to confirm that headers, sample rows, and key columns remain intact.
agent-sheet Setup
To integrate this tool into your Openclaw Skills environment, install the package via npm:
npm install -g agent-sheet@latest
Ensure your environment has the following dependencies available:
- python3
- awk
- sed
agent-sheet Data Schema & Taxonomy
agent-sheet organizes spreadsheet interaction through a structured taxonomy of identifiers and command outputs:
| Component | Description |
|---|---|
| entryId | The primary persistent identifier for a workbook in the local workspace. |
| unitId | A transient identifier used for active workbook sessions. |
| Range String | Standard A1 notation (e.g., 'Sheet1!A1:B10'), quoted for shell safety. |
| Tables | A1-anchored data structures with explicit header semantics. |
| Exports | Data can be streamed to stdout for pipelines or saved to specific file paths using --to-file. |
agent-sheet Advanced Features
- Script JS Fallback: Execute bounded JavaScript snippets for complex formatting or layout changes that standard CLI commands cannot express.
- Shell Roundtrip Integration: Seamlessly stream workbook data through awk and sed before writing back verified results.
- Multi-Agent Handoff: Support for importing templates and checking anchors to ensure consistency across different agent task stages.
- Formula Analysis: Deep inspection of cell formulas to understand and replicate workbook logic programmatically.
- Anchored Review Tables: Automatically construct review tables with headers for human or agent-led verification.
SKILL.md
Loading
Related Openclaw Skills
Site Architecture
A comprehensive framework for planning intuitive website hierarchies, navigation systems, and SEO-optimized URL structures.
Podcast Marketing & SEO Skill
A comprehensive framework for planning, launching, and scaling podcasts through data-driven SEO, multi-channel distribution, and strategic repurposing.
Rustic Almanac Guide
A grounded, folksy daily briefing skill that translates real-time weather and seasonal shifts into practical life, work, and wellness advice.
Mountain Daoist Guide
A serene daily lifestyle assistant that blends real-time weather and calendar data with traditional Eastern wisdom to provide practical, metaphor-rich life guidance.
Word Cloud Generator
A python-powered skill for creating customizable word cloud images from various text sources including direct input, files, and entire directories.
Traffic Law Documents Parser
A specialized tool for parsing Chinese traffic accident legal documents and generating structured exam answers using Word COM integration.
Featured*
