Abstract
A method for translating a canonical AI skill instruction file into provider-specific tool definitions for multiple target AI frameworks while preserving behavioral constraint metadata through explicit mapping rules. The skill instruction file is authored in a single canonical format containing structured capability entries with execution parameters and behavioral constraints. A translation layer ingests this file and emits provider-specific adapter definitions for OpenAI function calling JSON schema, Anthropic Claude tool-use schema, LangChain Tool class Python stub, and Model Context Protocol capability manifest JSON. Constraints that have no direct equivalent in the target framework are emitted as a standardized annotation block rather than silently dropped. The mapping table governing all constraint translations is published openly so that the preservation rules are fully reproducible. The translation is bidirectional: provider-specific definitions can be parsed back into the canonical format when an authoritative mapping exists.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Burton, Aaron, "Cross-Provider Skill File Portability Shim for Behavioral Constraint Preservation Across AI Tool Frameworks", Technical Disclosure Commons, (April 24, 2026)
https://www.tdcommons.org/dpubs_series/9913