What problem does it solve?
Learned agent patterns are normally locked to a single project, forcing every new project or machine to start from zero. This Skill publishes patterns to IPFS via Pinata and lets any peer fetch and merge them, turning local learning into a shareable artifact.
Core Features & Use Cases
- Publish patterns to IPFS: Store the current project's learned patterns and receive a content-addressed CID to share with peers.
- Fetch and merge remote patterns: Load a peer's CID and merge patterns into local state, with conflicts resolved by recency.
- Mirror across projects: Read patterns from a sibling project on disk and republish them under a new CID, useful for monorepos or fleets of related projects.
- Use Case: Before starting a fresh project, fetch the parent project's pattern CID so the new project's agents begin with prior knowledge instead of cold-start learning.
Quick Start
Publish my current project's learned patterns to IPFS and give me the CID so I can share it with my other project.