shelv

What This Skill Does

Shelv is an advanced productivity tool for the OpenClaw AI agent that transforms monolithic PDF documents into granular, navigable, and search-ready Markdown filesystems. By automating the extraction of text, headers, and metadata, Shelv allows you to treat complex documents like contracts, research papers, or technical manuals as a collection of standard local files. Once hydrated into your workspace, these documents become accessible through native Unix commands like grep for searching specific clauses, find for locating content by name, or cat for quick reading. Shelv bridges the gap between unstructured visual PDFs and structured text-based data, enabling seamless integration with other developer workflows.

Installation

To integrate Shelv into your OpenClaw environment, execute the following command in your terminal: clawhub install openclaw/skills/skills/joshuaohanlon/shelv

Ensure you have your API key ready from shelv.dev. This key must be exported as an environment variable (SHELV_API_KEY) to allow the agent to authenticate against the Shelv API. The skill uses local helper scripts for orchestration, so ensure your OpenClaw environment supports standard shell script execution.

Use Cases

• Legal Discovery: Quickly parse massive legal contracts into individual markdown files to search for specific liability clauses or dates using grep.
• Academic Research: Deconstruct complex academic papers into modular components (abstract, methodology, data analysis) for easier study and reference.
• Documentation Management: Convert large product manuals or technical books into a searchable local archive to query within your coding environment.
• Data Extraction: Hydrate structured information from PDFs directly into your codebase for further manipulation with scripts.

Example Prompts

1. "Shelv the Q4_Agreement.pdf file in my downloads and let me know when I can grep for the termination clause."
2. "I need to search through the Research_Study_2024.pdf. Please upload it to Shelv, hydrate it, and find every mention of 'p-value' using the filesystem."
3. "Upload the Manual.pdf to a new shelf using the 'book' template and confirm when the filesystem is ready for me to explore."

Tips & Limitations

• Review Mode: Use the --review flag for sensitive documents to verify the structure before final hydration.
• Templates: Use specific templates like legal-contract or academic-paper to improve the quality of the generated markdown structure; auto-detection is powerful but templates ensure consistency.
• Asynchronous Processing: Large PDFs may take several minutes to process. Always use the shelv-poll-status.sh script to verify when a shelf reaches the 'ready' state before attempting to hydrate files.
• Workspace Cleanliness: Remember that hydration creates directories in ~/.openclaw/workspace/shelves/. Use standard file commands to clean up old shelves that are no longer needed to save disk space.