hierarchical-llm-agents-acting-gridworld-cooperation-tasks-overcooked-layouts CLM-0070In hierarchical LLM agents acting in gridworld cooperation tasks (Overcooked layouts) with unseen partners, supplying an accurate partner-skill estimate as planner prompt context does not by itself stop the agent from continuing a stale, non-complementary skill; the estimate only changes behaviour when it is wired into an explicit interrupt rule that aborts the in-progress skill when new partner actions are unlikely under the committed estimate.
Capability: Turning plans into physical action
Observed on
Hierarchical LLM planners executing temporally extended skills alongside a partner whose strategy shifts mid-episode; holds in loosely coupled layouts (Open, Ring) and weakly in ti.
Sources
- Measured across 3 layouts x 3 partner groups (12 behaviour-preference partners, 5 seeds), GPT-4o and GPT-5.2. Skill-prediction accuracy was nearly identical to ProAgent (0.79 vs 0.78 Open) while the belief-action gap fell from 0.41 to 0.20 (Open) and 0.38 to 0.28 (Ring). An ablation with posterior-as-control-only beat posterior-as-prompt-only in most settings, and contradiction-gating matched heuristic triggers' reward with roughly 2 replans per episode versus tens to over a hundred. [truncated]
Disagreeing is the most useful thing you can do here. Nothing on this page waits for a reviewer: the counts above move when evidence lands, and every source stays attached with who brought it.
Related claims
- In LLM multi-agent systems that reflect after failure, restricting reflection to the single agent responsible for the first incorrect step — located by a trained step-level correctness scorer and paired with a counterfactually generated corrected step — raises task success more than having every agent reflect on the trajectory, because reflections written by agents that behaved correctly enter their memory as misleading guidance for later tasks.Getting more from several models than from one · reviewed by AI
- For LLM agents acting in partially observable environments, replacing the raw action-observation log in the prompt with an externally maintained Bayesian posterior over latent states makes the agent's action choice a function of the belief alone, removing surface-form sensitivity in which two histories with identical posteriors but different serialized text elicit different actions.Planning a sequence of actions before acting · reviewed by AI
- When automatically rewriting the prompts of a multi-agent LLM system from its failures, locating the responsible agent by injecting a hint into one agent at a time and using that agent's hint-corrected intermediate output as the supervision signal produces prompt updates that generalize better than critique derived only from the system-level output.Improving its own scaffolding from failures · reviewed by AI
- In multi-agent LLM systems where one agent starts with an erroneous shared belief, broadcasting every agent's full context at every step raises the rate of false assertions above doing no synchronization at all, because the error propagates to agents that were previously correct — and the harm appears only in tasks where one wrong fact cascades across semantically linked dimensions (destination to airport to weather), not where agent contexts are largely orthogonal.Stating false facts confidently · reviewed by AI
- For long-horizon embodied task planning where a symbolic model defines action preconditions and effects, correcting a violated plan by edits the action model itself implies (inserting the missing navigate or open) raises plan success more per LLM call than returning the violation to the LLM for replanning, and the gain persists for frontier models as well as compact ones.Planning a sequence of actions before acting · reviewed by AI
Notes
Drafted from the paper by a model and filed unreviewed. Visible here so it can be read, not because anyone has vouched for it: it does not move any technique's standing and does not count toward the internal scorecard. Drafted confidence: medium. Falsifier as drafted: An agent given the same posterior only in its prompt, with no interrupt mechanism, closes the belief-action gap as much as the interrupt-gated version at matched replan counts. Proposed technique, not catalogued: Contradiction-gated replanning from a Bayesian partner-skill posterior.