# Risk-Sensitive Agent Compositions

arXiv: https://arxiv.org/abs/2506.04632 · HTML: https://mlsanity.com/paper/2506.04632/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Agent threats, guardrails and governance, one of 16 directions in LLM agents and tool use.
- Builds on 26 of its 37 references that are in the map.
- First posted on arXiv 5 Jun 2025.

## Placed in

- LLM agents and tool use / [Agent threats, guardrails and governance](https://mlsanity.com/agents/agent-threats-guardrails-and-governance/)

## Builds on (most cited first, 26 of 26)

- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [Addressing Function Approximation Error in Actor-Critic Methods](https://mlsanity.com/paper/1802.09477/) (2018)
- [Do As I Can, Not As I Say: Grounding Language in Robotic Affordances](https://mlsanity.com/paper/2204.01691/) (2022)
- [CoT-VLA: Visual Chain-of-Thought Reasoning for Vision-Language-Action Models](https://mlsanity.com/paper/2503.22020/) (2025)
- [AFlow: Automating Agentic Workflow Generation](https://mlsanity.com/paper/2410.10762/) (2024)
- [Minigrid & Miniworld: Modular & Customizable Reinforcement Learning Environments for Goal-Oriented Tasks](https://mlsanity.com/paper/2306.13831/) (2023)
- [Multi-Goal Reinforcement Learning: Challenging Robotics Environments and Request for Research](https://mlsanity.com/paper/1802.09464/) (2018)
- [Agentic AI for Scientific Discovery: A Survey of Progress, Challenges, and Future Directions](https://mlsanity.com/paper/2503.08979/) (2025)
- [Compositional Reinforcement Learning from Logical Specifications](https://mlsanity.com/paper/2106.13906/) (2021)
- [Plan-Seq-Learn: Language Model Guided RL for Solving Long Horizon Robotics Tasks](https://mlsanity.com/paper/2405.01534/) (2024)
- [Guiding Long-Horizon Task and Motion Planning with Vision Language Models](https://mlsanity.com/paper/2410.02193/) (2024)
- [Efficient Risk-Averse Reinforcement Learning](https://mlsanity.com/paper/2205.05138/) (2022)
- [Reflective Planning: Vision-Language Models for Multi-Stage Long-Horizon Robotic Manipulation](https://mlsanity.com/paper/2502.16707/) (2025)
- [Flow: Modularized Agentic Workflow Automation](https://mlsanity.com/paper/2501.07834/) (2025)
- [Near-Minimax-Optimal Risk-Sensitive Reinforcement Learning with CVaR](https://mlsanity.com/paper/2302.03201/) (2023)
- [Regret Bounds for Risk-Sensitive Reinforcement Learning](https://mlsanity.com/paper/2210.05650/) (2022)
- [QualityFlow: An Agentic Workflow for Program Synthesis Controlled by LLM Quality Checks](https://mlsanity.com/paper/2501.17167/) (2025)
- [Risk-Sensitive Sequential Action Control with Multi-Modal Human Trajectory Forecasting for Safe Crowd-Robot Interaction](https://mlsanity.com/paper/2009.05702/) (2020)
- [Risk-Conditioned Distributional Soft Actor-Critic for Risk-Sensitive Navigation](https://mlsanity.com/paper/2104.03111/) (2021)
- [Risk-averse autonomous systems: A brief history and recent developments from the perspective of optimal control](https://mlsanity.com/paper/2109.08947/) (2021)
- [SPIRE: Synergistic Planning, Imitation, and Reinforcement Learning for Long-Horizon Manipulation](https://mlsanity.com/paper/2410.18065/) (2024)
- [Value-Informed Skill Chaining for Policy Learning of Long-Horizon Tasks with Surgical Robot](https://mlsanity.com/paper/2307.16503/) (2023)
- [Risk-Averse Stochastic Shortest Path Planning](https://mlsanity.com/paper/2103.14727/) (2021)
- [A Survey of Risk-Aware Multi-Armed Bandits](https://mlsanity.com/paper/2205.05843/) (2022)
- [Conditional Value-at-Risk for Reachability and Mean Payoff in Markov Decision Processes](https://mlsanity.com/paper/1805.02946/) (2018)
- [Risk-Aware Algorithms for Combinatorial Semi-Bandits](https://mlsanity.com/paper/2112.01141/) (2021)

## Built on by (most cited first, 1 of 1)

- [Spawn Freely, Act Sparingly: Progressive Risk Vesting for Recursive LLM-Agent Trees](https://mlsanity.com/paper/2609.01035/) (2026)

## Abstract (arXiv)

From software development to robot control, modern agentic systems decompose complex objectives into a sequence of subtasks and choose a set of specialized AI agents to complete them. We formalize agentic workflows as directed acyclic graphs, called agent graphs, where edges represent AI agents and paths correspond to feasible compositions of agents. Real-world deployment requires selecting agent compositions that not only maximize task success but also minimize violations of safety, fairness, and privacy requirements which demands a careful analysis of the low-probability (tail) behaviors of compositions of agents. In this work, we consider risk minimization over the set of feasible agent compositions and seek to minimize the value-at-risk and the conditional value-at-risk of the loss distribution of the agent composition where the loss quantifies violations of these requirements. We introduce an efficient algorithm which traverses the agent graph and finds a near-optimal composition of agents. It uses a dynamic programming approach to approximate the value-at-risk of agent compositions by exploiting a union bound. Furthermore, we prove that the approximation is near-optimal asymptotically for a broad class of practical loss functions. We also show how our algorithm can be used to approximate the conditional value-at-risk as a byproduct. To evaluate our framework, we consider a suite of video game-like control benchmarks that require composing several agents trained with reinforcement learning and demonstrate our algorithm's effectiveness in approximating the value-at-risk and identifying the optimal agent composition.