Abstract
A method for discovering and ranking skill instruction files published to an AI agent marketplace using a composite scoring function applied to a vector index. When a querying agent submits a natural language query expressing a desired capability outcome, the system computes a composite relevance score for each indexed skill by combining embedding cosine similarity between the query and the skill description, declared constraint compatibility between the skill and the querying agent's operational profile, and an optional historical quality signal derived from prior benchmark completions. The ranking layer operates entirely upstream of any licensing, payment, or entitlement system and returns a plain ordered list of candidate skill identifiers with scores, performing no access control of any kind.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Burton, Aaron, "Semantic Discovery and Composite Ranking of Skill Instruction Files in an AI Agent Marketplace", Technical Disclosure Commons, (April 24, 2026)
https://www.tdcommons.org/dpubs_series/9915