Abstract
Proposed herein is a decision-map-driven Troubleshooting Retrieval and Context Software Development Kit (SDK) that converts an incident into an abstract, structured representation and deterministically routes retrieval, using supervisory-tagged data, across non-linear knowledge sources. The sources may include historical incidents, methods of procedure (MOPs), workflows, documentation, telemetry, and topology or Configuration Management Database (CMDB) relationships. The SDK assembles a pre-tokenization structured context bundle that bounds and standardizes evidence for downstream reasoning, improving precision and consistency relative to static runbooks or generic retrieval-augmented generation (RAG). In some implementations, the SDK enriches incident cases with normalized public outage and environmental intelligence based on site or location metadata to improve branching and attribution.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
P C, Nikhil; Jeuk, Sebastian; Khan, Atahar; and Nair, Shibu, "ADAPTIVE MULTI-AGENT ARTIFICIAL INTELLIGENCE FOR INFORMATION TECHNOLOGY OPERATIONS SYSTEM FOR DETERMINISTIC, TOPOLOGY-AWARE NETWORK TROUBLESHOOTING AND REMEDIATION WITH CLOSED-LOOP LEARNING", Technical Disclosure Commons, (July 28, 2026)
https://www.tdcommons.org/dpubs_series/11155