ContractEval reveals hidden failures in AI procedural task completion
ContractEval: Query-Conditioned Execution Matching for Procedural Instruction Conformance
Artificial Intelligence
Summary
When AI systems follow step-by-step instructions, sometimes they seem to do things right but actually skip important steps or checks. The authors created ContractEval to spot these hidden mistakes by making clear which rules should apply to the AI’s task and checking if they really did. This method can find errors that other ways miss, like doing things out of order or breaking key rules. ContractEval helps people audit AI behaviors instead of just guessing if the final answer looks okay.
What this means in practice
- •For software quality assurance teams: Detect when AI-driven software misses required procedural steps or violates operational rules during automated processes.
- •For robotics control engineers: Audit robot task executions to identify skipped checks or wrong actions that could cause failures despite acceptable outputs.
Authors
Praphul Singh, Shanu Kumar, Akshat Agarwal, Ganesh Kumar
Abstract
As LLM agents move from answering questions to carrying out procedures, failures can be unwarranted rather than visibly wrong: the final response looks acceptable even though the system skipped the check, branch, dependency, or invariant that made the answer justified. Output-only evaluation sees the answer, and trace-aware judging sees activity, but neither identifies which obligations were active for the query. We introduce CONTRACTEVAL, a diagnostic framework for making those active obligations explicit. It represents procedural instructions as query-active obligations and matches them against response or trace evidence, turning omissions, wrong branches, ordering errors, extra actions, invariant breaches, and output-contract violations into distinct conformance failures. On a controlled suite of audited procedural contracts, output-only and trace-aware LLM judges miss many injected structural failures; under gold expected and observed graphs, ContractEval detects and localizes all of them. LLM-backed extraction preserves much of this signal but remains calibration-sensitive. ContractEval is therefore not a compliance guarantee; it makes procedural conformance auditable rather than implicit in final-answer quality.