What problem does it solve?
Capturing what makes a long-lived LingTai agent network distinctive and packaging it for others is error-prone: agents confuse the network's distilled culture with the recipe that originally seeded it, writes silently fail, and bundles fail validation. This Skill guides the full export workflow so the resulting recipe bundle is valid, safe to publish, and faithful to how the network actually behaves.
Core Features & Use Cases
- Scope Disambiguation: Distinguishes the three recipe-shaped artifacts (inner-network distillation, the project's own
.recipe/, and the TUI snapshot) before authoring anything.
- Human-in-the-Loop Authoring: Collects recipe id, name, description, layers, library, and languages via email, then writes
recipe.json, greet.md, comment.md, and optional library skills in one pass.
- Validation and Safe Publishing: Runs
validate_recipe.py, performs a sensitivity sweep for private names or unreleased details, then initializes git and optionally pushes to GitHub.
- Use Case: After weeks of running a research network, distill its delegation rules, communication norms, and custom skills into a
scholar-distiller bundle that colleagues can clone and apply with /setup.
Quick Start
Export this network's culture as a recipe bundle named scholar-distiller, validate it, and prepare it for publishing to GitHub.
