GIST-CMTF: Goal-State Inference for Causal Minimal Tool Filtering in LLM Agents
2026-06-15 • Artificial Intelligence
Artificial Intelligence
AI summaryⓘ
The authors address a problem where AI agents using tools sometimes pick the wrong goal because user requests can match multiple possible intentions. They improve on a method called Causal Minimal Tool Filtering by adding a layer (GIST-CMTF) that guesses which goals might fit the request and checks if it's unclear. When there's uncertainty, the system asks for clarification before choosing tools, leading to fewer mistakes. Their experiments show that this approach is much better at completing tasks correctly and reduces wrong-goal actions, while still keeping tool choices simple and efficient.
Large Language Models (LLMs)Tool-augmented agentsCausal Minimal Tool Filtering (CMTF)Goal-state inferenceState-transition vocabularyTask success rateWrong-goal executionClarification actionsTool relevanceSymbolic goals
Authors
Rahul Suresh Babu, Rohit Shukla
Abstract
Tool-augmented LLM agents rely on runtime filtering to decide which tools should be visible at each step. Causal Minimal Tool Filtering (CMTF) reduces tool-choice confusion by exposing only the next causally necessary tool frontier, but it assumes that the user request has already been mapped to a symbolic goal state. In practice, requests such as "handle my appointment" or "take care of this email" may correspond to multiple possible goals. This creates wrong-goal execution, where an agent follows a valid causal tool path for an unintended objective. We introduce GIST-CMTF, a goal-state inference layer that predicts candidate symbolic goals over the same state-transition vocabulary used by CMTF, estimates ambiguity, and either applies CMTF or exposes clarification as a causal action that produces missing goal or state variables. We evaluate GIST-CMTF across seven model backends, six filtering methods, and 120 controlled tool-use tasks. GIST-CMTF achieves 97.0% task success, compared with 80.1% for top-goal CMTF and 82.9% for semantic-goal CMTF. It reduces wrong-goal execution from 19.4% under top-goal CMTF to 2.5%, while preserving the one-tool exposure of causal filtering and using substantially fewer tokens than all-tools exposure. These results suggest that reliable tool-augmented agents should validate goal state, not only tool relevance, before exposing external actions.