[docs] Add AgentCore RL Toolkit to ecosystem list

README.md

@@ -140,6 +140,10 @@ These are not just demos. They are independent systems that use slime as a reusa
 
 [**qqr**](https://github.com/Alibaba-NLP/qqr) (a.k.a. hilichurl) is a lightweight extension for slime designed to evolve open-ended agents. It implements the **ArenaRL** algorithm to tackle discriminative collapse through tournament-based relative ranking (**e.g., Seeded Single-Elimination, Round-Robin**) and seamlessly integrates the **Model Context Protocol (MCP)**. qqr leverages slime's high-throughput training capabilities to enable scalable, distributed evolution of agents in standardized, decoupled tool environments.
 
+### ☁️ ART: Scalable and Sandboxed Agentic RL on AWS Bedrock AgentCore Runtime
+
+[**ART (AgentCore RL Toolkit)**](https://github.com/awslabs/agentcore-rl-toolkit) is an SDK that adapts production agents for RL training on **AWS Bedrock AgentCore Runtime**. AgentCore Runtime provides auto-scaled and sandboxed agent execution environments well-suited for running many parallel agent rollouts securely. Using ART, user only needs to apply a decorator (`@app.rollout_entrypoint`) to their agent codes for RL adaption while the same production agent harness is reused directly, where token capture for RL is handled at model gateway layer. ART uses slime as one option of training backends, enabling users to easily optimizing the production agent model with RL training algorithms in slime.
+
 Together, these projects show the main idea behind slime: one high-performance RL kernel can support frontier model post-training, online agent optimization, verifiable environments, omni-modal rollouts, kernel-generation agents, and rollout-system research without changing the core training loop.
 
 ## Arguments Walkthrough

README_zh.md

@@ -142,6 +142,10 @@ slime 被当作 RL 基础设施来开发，因为“脚本能跑起来”远远
 
 [**qqr**](https://github.com/Alibaba-NLP/qqr) (a.k.a. hilichurl) 是一个用于演化 open-ended agent 的 slime lightweight extension。它实现 **ArenaRL** algorithm，通过 tournament-based relative ranking（例如 Seeded Single-Elimination、Round-Robin）缓解 discriminative collapse，并无缝集成 **Model Context Protocol (MCP)**。qqr 利用 slime 的高吞吐训练能力，在标准化、解耦的 tool environment 中实现可扩展的分布式 agent evolution。
 
+### ☁️ ART: Scalable and Sandboxed Agentic RL on AWS Bedrock AgentCore Runtime
+
+[**ART (AgentCore RL Toolkit)**](https://github.com/awslabs/agentcore-rl-toolkit) 是一个能够将真实生产环境中的 agent 适配到 **AWS Bedrock AgentCore Runtime** 上进行 RL 训练的工具包。AgentCore Runtime 提供了能够自动扩展以及沙盒式封闭管理的智能体运行环境，这非常适合安全地并行运行大量 agent rollouts。利用 ART，用户只需在 agent 代码上使用一个 decorator（`@app.rollout_entrypoint`），即可在直接复用生产环境的 agent harness 基础上完成 RL 训练的适配，其中用于 RL 训练的 token capture 则在 model gateway layer 中完成。ART 将 slime 列为 RL 训练的后端选项之一，帮助用户能够轻松地使用 slime 中的 RL 训练算法优化生产环境上的 agent 模型。
+
 这些项目共同体现了 slime 的核心思路：一个高性能 RL kernel 可以同时支撑 frontier model post-training、online agent optimization、verifiable environment、omni-modal rollout、kernel-generation agent 和 rollout-system research，而不需要改变核心 training loop。
 
 ## 参数说明