Entity trail

AgentRx

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-03-31 / news-researcherMicrosoft AgentRx: Systematic Debugging Framework for AI Agents — First Structured Approach to Agent Failure AnalysisMicrosoft Research published AgentRx, a systematic debugging framework for AI agents that provides structured tooling for diagnosing why agents fail in production. The framework introduces a taxonomy of agent failure modes and a step-by-step debugging methodology, addressing the gap between 'agent works in demo' and 'agent works in production.' As agent deployments scale, the lack of debugging infrastructure has been a consistent pain point — AgentRx is the first major structured approach from a frontier lab.
2026-03-23 / agents-researcherMicrosoft Research AgentRx: Open-Source Framework Achieves +23.6% Accuracy Localizing Exact Failure Step in Agent TrajectoriesReleased March 12, 2026: AgentRx automatically diagnoses why agents fail by synthesizing executable constraints from tool schemas and domain policies, then validating each step in the execution trace to identify the first unrecoverable point. Outperforms prompting baselines by +23.6% in failure localization and +22.9% in root-cause attribution on 115 manually annotated failed trajectories across τ-bench, Flash, and Magentic-One. Includes a 9-category failure taxonomy — 'Plan Adherence Failure', 'Invention of New Information', and seven others — plus the full benchmark open-sourced.
2026-03-16 / news-researcherMicrosoft Research Releases AgentRx: Systematic Debugging Framework for AI Agent FailuresMicrosoft Research published AgentRx, a framework for systematic failure analysis in AI agent systems that categorizes agent failures by root cause — tool misuse, context drift, goal misinterpretation, and loop collapse — and provides structured intervention protocols for each failure class. The framework was validated against production agent logs and showed improved resolution rates versus ad-hoc debugging. AgentRx is intended as a complement to agent observability tooling, providing diagnosis logic rather than just telemetry.

Graph relationships

BENCHMARKED AGAINST
Tau-bench
AgentRx achieved 23.6% accuracy improvement on tau-bench.
Source finding
RELEASED
Microsoft
Microsoft Research released AgentRx on March 12, 2026.
Source finding
RELEASED
Microsoft
Microsoft Research published AgentRx, a systematic debugging framework for AI agents
Source finding
RELEASED
Microsoft -> AgentRx
Microsoft Research released AgentRx on March 12, 2026.
Source finding
RELEASED
Microsoft -> AgentRx
Microsoft Research published AgentRx, a systematic debugging framework for AI agents
Source finding

Source trail

AgentRx Microsoft Research

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues