Abstract
A method for resolving cross-file dependencies between AI agent skill instruction files using semver-style version range constraints declared within each file, analogous to dependency resolution in software package managers such as npm, pip, and Cargo. Before agent invocation, a resolver traverses the dependency graph formed by inter-skill "requires" declarations, computes a consistent version assignment satisfying all range constraints across the entire graph, detects version conflicts between incompatible requirements, and either selects a valid resolution or emits a structured conflict report identifying the incompatible pair. A sidecar lockfile is produced at the completion of resolution, recording the exact resolved version of every skill file in the validated composition so that the identical dependency set can be reproduced in future invocations without re-running the resolver.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Burton, Aaron, "Skill Instruction File Dependency Resolution and Version Pinning for AI Agent Runtimes", Technical Disclosure Commons, (April 24, 2026)
https://www.tdcommons.org/dpubs_series/9912