expertpack-export

What This Skill Does

The expertpack-export skill provides a robust framework for converting the accumulated knowledge, personality, and operational logic of an OpenClaw AI agent into an ExpertPack composite. An ExpertPack is a standardized, portable knowledge format that allows you to move agent intelligence across different environments or back up your agent's state in a structured, human-readable way.

This tool operates through a six-stage pipeline: Scanning, Proposal, Distillation, Composition, Validation, and Review. It automatically maps disparate files within your workspace into categorized packs (agent, person, product, or process), extracts core knowledge assertions, scrubs sensitive credentials like API keys to ensure security, and packages the data into valid ExpertPack schemas. By utilizing this skill, you ensure your agent's development history and operational context remain organized and transferable.

Installation

To add this capability to your agent, run the following command in your terminal or command-line interface:

clawhub install openclaw/skills/skills/brianhearn/expertpack-export

Ensure that you have appropriate write permissions to your local workspace, as the skill will generate an export/ directory containing your packaged knowledge files.

Use Cases

1. Knowledge Portability: Moving an agent's persona and logic from a local testing environment to a production deployment without losing core operational context.
2. Disaster Recovery & Archiving: Creating regular, version-controlled snapshots of your agent's knowledge base to restore service in the event of local workspace corruption.
3. Collaborative Knowledge Sharing: Exporting specific 'process' or 'product' packs to share internal logic with other team members or other OpenClaw instances.

Example Prompts

1. "Export my current agent state into an ExpertPack so I can migrate it to my server."
2. "Scan my current workspace and show me the proposed structure for an ExpertPack export before you begin distilling."
3. "Run a full export of the workspace and validate the output against the ExpertPack schema rules."

Tips & Limitations

• Verification: Always review the scan output generated by scan.py. If the agent suggests incorrect mappings, you can manually guide the process before distillation begins.
• Secrets Scrubbing: While the distill.py script automatically filters known API key patterns, always perform a final visual audit of your export files before uploading them to public repositories.
• Non-Destructive: This skill is strictly for exportation. It does not handle importing or hydrating a new agent from an existing pack; you will need to utilize the OpenClaw import tools for that purpose.
• Prerequisites: Familiarize yourself with the ExpertPack schema documentation to better understand how your agent's data is being structured, especially if you intend to modify the output files for downstream applications.