SAGE: Symbolic Action-Gating and Editing for LLM Task Planners

Trung Minh Bui; JongSul Moon; YoungOuk Kim; Quang-Ngoc Phung; Se-Woong Jun; Dongin Shin

cs.ROSep 28, 2026

SAGE: Symbolic Action-Gating and Editing for LLM Task Planners

Authors: Trung Minh Bui, JongSul Moon, YoungOuk Kim, Quang-Ngoc Phung, Se-Woong Jun, Dongin Shin

Organizations: Intelligent Robotics Research Center, Korea Electronics Technology Institute (KETI), Seongnam 13449, Republic of Korea

Abstract

Large language models (LLMs) are now the default cognitive core of embodied household agents, yet the plans they emit are rarely checked against a grounded model of the environment before execution, and the task-success they report is often measured on benchmarks so saturated that no method can be separated from another. We present SAGE (Symbolic Action-Gating and Editing), a single-LLM planner built from two lightweight mechanisms: a domain-agnostic symbolic gate (~250 lines of Python, zero tokens, $O(|π|)$ ) that blocks precondition-violating actions with typed reasons as a runtime safety monitor, and a local edit that regenerates only the failed sub-goal's suffix, keeping completed and untouched work intact; a hybrid seed+live memory store supports cold-start coverage. We evaluate under a leak-free protocol (leave-one-out retrieval) over five open-weight models and a 75-task AI2-THOR benchmark. On the standard benchmark goal-completeness saturates (52% of instances trivially solved) and SAGE ties strong hierarchical baselines. On a harder, method-agnostic multi-goal composition, SAGE's completeness lead re-emerges large (+0.06 to +0.23 across four models). Under injected mid-execution failures, SAGE recovers as reliably as whole-plan replanners at 2.4-3.3x fewer LLM calls. As a verify-before-execute gate, the symbolic monitor blocks unsafe actions before actuation and raises simulator-reported step-success for every planner tested (up to +0.11), a signal the verifier never sees (non-circular). Because the gate calls no model (0.008 ms/plan), it is a safety layer that runs essentially free on the edge: SAGE planning reproduces its quality on a Jetson AGX Orin, where small-model verification helps most. We release the benchmark, the leak-free protocol, the recovery and safety-gate harnesses, and a verifier-portability study (auto-induced on ALFWorld, 0.89 held-out).

Figures & tables

Fig. 1: SAGE gates and edits. A naive plan Picks the potato before the fridge is open; the zero-token symbolic gate flags the precondition violation and the sub-goal–local edit regenerates only the failed sub-goal’s suffix, yielding a certified plan without a whole-plan replan (frames: AI2-THOR). Bottom: under injected mid-execution failures (all methods recover $100\%$ ), SAGE recovers with $2.4$ – $3.1\times$ fewer LLM calls than whole-plan replanning (up to $3.3\times$ on compound tasks).

Fig. 2: SAGE pipeline. A task is hierarchically decomposed; each sub-goal is expanded with few-shot retrieval from hybrid memory, then every action passes a zero-token symbolic gate (verify-before-execute). A flagged precondition violation triggers a sub-goal-local edit that regenerates only the failed sub-goal’s suffix and re-enters the gate; gate-passing steps form the certified plan executed in AI2-THOR.

Model	Direct	Hier-FS	SAGE	$\Delta$	SAGE steps	SAGE p-valid
qwen2.5:7b	0.718	0.776	0.854	+0.078	12.0	0.943
qwen2.5:14b	0.757	0.783	0.902	+0.119	12.0	0.937
llama3.2	0.629	0.499	0.728	+0.229	9.0	0.933
mistral-nemo	0.570	0.824	0.880	+0.057	12.7	0.952

TABLE I: Goal completeness on the harder multi-goal compound benchmark (leak-free). The base-set completeness tie disappears: SAGE leads by a wide, consistent margin that grows as the base model weakens, while emitting shorter, more valid plans than the verbose hierarchical baseline.

Explore similar work

Jun 12, 2026cs.CL

SIMMER: Benchmarking Latent Failures in LLM Executable Planning with a World Model

Large language models (LLMs) are increasingly deployed as planners for autonomous agents in household environments. While existing benchmarks evaluate whether LLM-generated plans execute successfully, they overlook a critical type of failure: latent failures. Unlike immediate failures that trigger instant feedback at execution time and enable timely correction, latent failures do not immediately halt plan execution but silently compromise goal achievement. In severe cases, they cause irreversible harm. To address this gap, we introduce SIMMER, a benchmark for evaluating latent failures in LLM planning through a human-curated symbolic world model grounded in the kitchen domain. SIMMER defines a world model comprising 77 actions, 262 unique objects, and approximately 46,800 possible interactions that are semantically realistic, derived from real-world cooking scripts. It then leverages a state machine executor that validates plans against the world model and detects immediate precondition violations, latent hazards, and irreversible failures. Experiments across six LLMs show that even frontier models achieve at most 17% error-free plans. Moreover, up to 56% of plans contain latent failures, the majority of which lead to irreversible consequences. We further demonstrate that explicit state reasoning via counterfactual foresight simulation can reduce latent failures by up to 72% and irreversible cases by up to 75%, suggesting a promising direction for more robust LLM planners.

Xiaoxin Lu, Ranran Haoran Zhang, Rui Zhang

The Pennsylvania State University, State College, PA, USA

Large Language Model Planning Large Language Models Fail Latent Failure Patterns Latent World Models World Models Automata

Ask PDF

Sep 29, 2026cs.AI

Do LLM Agents Execute the Plans They Declare? From Planning-Mode Declaration to Pattern-Specific Execution

Large language models (LLMs) enable agents to solve long-horizon tasks by generating a plan and then executing it in an environment. However, successful planning requires two distinct capabilities: selecting an appropriate plan for the task and executing it faithfully. Existing planner--executor systems can fail at either stage, while final task success alone cannot distinguish selection from execution failures. We therefore study the Plan Declaration--Execution Gap and introduce Planning-as-Routing, where an LLM declares one of four planning modes: Predefined, Sequential, Hierarchical, or Search, and a deterministic router dispatches the task to the corresponding pattern-specific executor. Across four benchmarks and three LLMs, we find three consistent patterns. First, generic Plan+ReAct often fails to preserve declared planning structure, especially for longer plans: across three benchmarks, only (22)--(45%) of trajectories preserve it, whereas pattern-specific executors enforce the intended structure. Second, planning-mode effectiveness varies across environments and models: Search performs best on ALFWorld, Hierarchical on SWE-bench, and the strongest pattern can vary across models within the same benchmark. Third, the largest gains come from execution: pattern-specific executors improve task success from (0.48) to (0.92) on ALFWorld and from (0.36) to (0.44) on SWE-bench Verified over Plan+ReAct. Current LLMs, however, do not reliably select the strongest mode for each task, although few-shot examples improve selection in some benchmark--model combinations. Overall, reliable agent planning requires both effective mode selection and faithful execution: routing substantially closes the execution gap, while task-specific mode selection remains open.

Subba Reddy Oota, Francisco Herrera, Jordi Cabot Sagrera +2

ADIA Lab, Abu Dhabi, United Arab Emirates · University of Granada, Granada, Spain · Luxembourg Institute of Science and Technology, Luxembourg +2

Large Language Model Planning Large Language Model Agents Planning Long-Horizon Task Planning

Ask PDF

Jun 3, 2026cs.CL

Agent Planning Benchmark: A Diagnostic Framework for Planning Capabilities in LLM Agents

Planning is central to LLM agents: before acting, an agent must decompose goals, select tools, reason over constraints, and decide when a task is infeasible. Yet existing agent evaluations often report only end-to-end success, making it difficult to determine whether failures stem from planning or execution. We introduce Agent Planning Benchmark (APB), a planning-specific diagnostic benchmark with 4,209 multimodal cases across 22 domains and five settings, covering holistic planning, feedback-conditioned step-wise planning, and robustness under extraneous tools, broken tools, and unsolvable tasks. Across 12 MLLMs, APB reveals systematic weaknesses in long-horizon planning, tool-noise robustness, calibrated refusal, and inference-time refinement. We further validate APB on 200 ToolSandbox tasks and 200

τ^2

-bench tasks, where APB-guided refinement consistently improves plan correctness, plan grade, and downstream execution metrics across three representative models. APB thus serves as an upstream diagnostic complement to execution benchmarks. The APB benchmark and code are available in this URL.

Haoyu Sun, Wenxuan Wang, Mingyang Song +5

Tongji University · Shanghai AI Laboratory · Harbin Institute of Technology +5

Large Language Model Planning Agentic Benchmarks Large Language Model Agents Planning Diagnostic Benchmark Diagnostic Framework

Ask PDF