Summary
When companies use multiple AI agents to handle software tasks, they need to understand the exact behaviors programmed into their systems, not just general documentation. The authors developed a method using seven AI agents to pull out thousands of detailed rules from hundreds of code repositories. Because many rules look very similar but come from different parts of the system, simple search methods mix them up easily. To fix this, the authors created a system that organizes rules using information about where each rule came from before searching, making sure agents find the right rules for the right parts. This helped fix over 100 software problems and discover sources of recent errors in updates.
What this means in practice
- •For software developers: Identify and resolve specific software defects by querying precisely scoped behavioral logic from large codebases using structured metadata and artifact tracking.
- •For quality assurance teams: Attribute failures in software releases to exact rules and services responsible, enabling targeted testing and mitigation of regression issues.
- •For enterprise ai system builders: Integrate a scalable knowledge layer to organize and retrieve behavioral artifacts from multiple services for enhanced multi-agent coordination and debugging.$Commercial implications: Provides a commercial product feature for enterprises requiring clear, actionable mappings of AI agent behaviors across services to improve incident management and maintenance.
Authors
Garima Agrawal, Prasun Das, Akshay Sarvade, Priyanka L, Minisha N, Sravani Joshna, Hemath Manivannan, Lavanya Korada, Sambhrit Pattanayak, Srashti Gupta, Siddu R, Prasad Kalyanasundaram
Abstract
Enterprises are deploying agentic AI across the software development lifecycle, and they need the behavioral logic of the systems they act on, not stakeholder documentation. A seven-agent reverse-engineering pipeline recovered this logic, processing 241 repositories across 18 services into thousands of artifacts describing scope, journeys, rules and acceptance criteria. Recovery does not confer usability. Entities like offer, eligibility, and activation recur across services under a shared vocabulary, related in name but operationally distinct, so one end-to-end journey implicates over a thousand near-identical rules that must not be conflated. Flat vector retrieval fails, returning an identical rule from the wrong service at high similarity with no signal of error. Content-level ontology, the traditional remedy, did not exist, and relating every occurrence of recurring entities would have cost more than the engagement and not survived a release cycle. We place the ontology at the artifact level. In this machine-generated corpus, each artifact retains its producer, component, service, domain and type, provenance that disambiguates before content is read. We present a knowledge layer that organizes before retrieving: a self-describing catalog, a sparse metadata registry, and a navigation hypergraph defined over artifacts rather than extracted entities. Queries resolve to artifact identifiers that constrain vector search, so the competing rule from another service is absent rather than outranked, making disambiguation structural, not statistical. Built in under a week and admitting new services by convention, the layer serves SDLC agents, developers, quality engineers and analysts. It supported root-cause analysis and resolution of 100+ production defects with developer verification and attributed 35 release regression failures to specific rules and services.