Executable synthetic environments provide scalable training tasks and verifiable feedback for long-horizon agents. Existing environment synthesis methods primarily generate tasks within individual environments. However, these tasks do not adequately capture real-world interactions across services, where actions in one service depend on information obtained from another. Synthesizing such tasks is challenging because data generated across environments must remain consistent to support executable and verifiable interactions. To address this challenge, we introduce CrossEnv, a framework that uses an entity-tool graph to identify connections between environments and aligns task-specific data to synthesize executable, verifiable cross-environment tasks. We further propose CrossEnv-Curriculum, which progresses from single-environment tool use to coordination across more environments and business domains, before introducing user interaction. Across four external benchmarks, combining cross-environment training data with this curriculum achieves competitive performance with fewer training examples. Further analyses support the benefits of cross-environment data and curriculum training, and provide evidence of transfer to new environment combinations, external tasks less similar to the training data, and longer workflows.
Figures & tables
Figure 1: Independent tasks and a cross-environment task. (a) Flight, hotel, and dining tasks are completed and verified independently. (b) A travel-planning task uses flight arrival time, hotel location, and personal preferences to guide restaurant reservation, with joint verification of cross-environment requirements.
Table 1: Environment and task synthesis capabilities. Exec.: real-service or code execution; Verif.: programmatic outcome checks; Ctrl.: adjustable complexity; Cross: multi-environment tasks; Consist.: cross-environment data alignment; Ext.: new-environment integration into joint tasks. : supported; : limited; : not explicitly provided.
Figure 2: Overview of the CrossEnv task synthesis framework. Entity–tool graph composition forms task skeletons, while tool synthesis, data alignment, and reference execution produce executable tasks with verifiable outcomes. Tasks are validated through reference replay and independent agent attempts.
Level
Stage
Train
Val.
Avg. Turns
L1
SFT
800
40
4.73
L2
RL
579
40
4.84
L3
RL
620
50
8.70
L4
RL
600
50
12.66
L5
RL + User
355
40
13.76
Table 2: CrossEnv training and validation data by curriculum level. Turns count tool calls and, for L5, one required clarification question.
Table 3: Performance on four external benchmarks. Source: hosted API; public checkpoint deployed locally; trained in this work. Avg. is weighted by subset size. Bold and underlined scores indicate the best and second-best among the displayed 1.7B–8B models.
Figure 3: Qwen3-8B capability scores from Base to L5. Shading highlights the corresponding training stages. Panel (e) reports a composite interaction score, not a success rate (Appendix F.2 ).
Figure 4: Curriculum vs. Mixed (8B): (a) strict success at similar GPU-hours; (b) cumulative rollouts.
Model
Pair
Success
Obs.
State
Answer
1.7B
Seen
+6.7
+40.7
+37.5
+22.2
Unseen
+5.3
+38.9
+ 0.0
+11.1
4B
Seen
+30.0
+43.7
+48.5
+32.2
Unseen
+24.0
+33.7
+26.9
+22.5
8B
Seen
+28.0
+44.3
+30.4
+26.4
Unseen
+16.0
+34.3
+14.6
+13.0
Table 4: CrossEnv gains over Base on seen and unseen environment combinations (percentage points). Obs.: observation.
Figure 5: Generalization to longer workflows. Scores show the average proportion of task checks passed (%).
Figure 6: Task similarity and CrossEnv gains over Base at 4B (left) and 8B (right). Lines show mean gains in four similarity groups with 95% confidence intervals.
Metric
Base
SFT
SingleEnv
CrossEnv
Benchmark performance
τ2 -bench
11.5
8.5
8.8
11.5
DeepPlanning
1.7
3.3
6.7
7.5
Claw-Eval
14.4
16.9
15.4
20.0
Execution capabilities
Output reuse
6.7
8.3
1.7
11.7
Table 5: Qwen3-8B comparison (%). Bold: row best.
Figure 7: Output yield and average environment-build cost. *: estimates.
Appendix figures & tables3 assets
Supplementary material from the paper’s appendix.
Appendix
Business domain ( C1 )
Capability categories ( C2 )
Travel
Flight booking; Accommodation booking; Food and dining; Transit and navigation; Earth science and environmental monitoring; Maps and location services; Weather and forecasting; Itinerary planning; Currency exchange; Messaging and notification; Task and project management; Local search; Calendar and scheduling.
Work & Office
Decision support; Messaging and notification; Task and project management; Calendar and scheduling; Email management; Document processing; Presentation and design; Knowledge management; Web search; Customer support.
Life Services
Food and dining; Transit and navigation; Earth science and environmental monitoring; Maps and location services; Weather and forecasting; Messaging and notification; Calendar and scheduling; Shopping; Home services; Local news; Medical appointment booking; Billing and payments; Arts, culture, and museums; Events and entertainment; Gaming and interactive experiences; Real estate services.
Finance & Investment
Currency exchange; Billing and payments; Crypto market data; Market intelligence; Stock and equity investment; Blockchain infrastructure and wallets; DeFi and DEX protocols; Accounting; Risk management and compliance.
Health & Wellness
Medical appointment booking; Biomedical research; Medical information lookup; Fitness tracking; Nutrition and diet; Mental wellness.
Academic Research
Earth science and environmental monitoring; Knowledge management; Biomedical research; Literature search; Citation analysis; Data collection and processing; Arithmetic and general calculation; Scientific and symbolic computing; Statistical analysis; Academic writing and collaboration; Data visualization; Experiment design.
Appendix
Table 6: Complete environment taxonomy. The table covers all 10 business domains and all 84 unique capability categories. Categories shared across domains are repeated with identical names; the number of domain–capability associations therefore exceeds 84.
Parameter
L1 SFT
L2–L5 RL
Learning rate
5×10−7 (peak)
5×10−7 (constant)
LR schedule
≈ 10% warmup, cosine decay
Constant
Global batch
8 assistant responses
16 task slots
Samples per task
Reference supervision
8 trajectories
PPO epochs
—
1
KL coefficient
—
0.001 in the loss
Appendix
Table 7: Training configuration for L1 SFT and L2–L5 RL. Each RL batch contains 128 sampled trajectories.
Model
Version / source
Frontier closed-source models
GPT-5.6-Luna
OpenAI model documentation API ID: gpt-5.6-luna
Claude-Haiku-4.5
Anthropic model announcement Claw-Eval API ID: bedrock/claude-haiku-4_5
Gemini-3.8-Flash
Google DeepMind model card API ID: gemini-3.8-flash
Qwen3.8-Max
Claw-Eval API ID: qwen3.8-max-0902 Qwen/Qwen3.8-2.4T-A95B (related model card)
Agents learn to act through interaction with environments, yet the environments used for training are often manually constructed or synthesized around predefined tasks and benchmarks. This task-centric paradigm makes it difficult to scale environments that reflect realistic and evolving workflows where diverse tasks can naturally emerge from the underlying world. We introduce AgentMercury, a scalable framework for synthesizing executable environments from high-level business scenarios. Rather than constructing an environment for a specific task, AgentMercury first instantiates a persistent world with entities, services, tools, state, and executable cross-service invariants, from which diverse tasks and interaction trajectories can subsequently emerge. We construct 4,783 executable environments spanning 14 industries and 50 countries, and use them as training substrates for reinforcement learning. Despite being generated without targeting the evaluation benchmarks, policies trained on these business-oriented environments improve substantially on both enterprise workflows and out-of-domain benchmarks spanning reasoning, coding, scientific computing, and tool use. In our experiments, Qwen3.5-4B improves from 12.3 to 15.7 on EnterpriseOps-GYM and from 45.9 to 56.0 on AIME26 after training on AgentMercury environments. We further show that the construction process itself can be learned: fine-tuning Qwen3.5-35B-A3B on construction traces increases executable-world authoring success from 3.3% to 83.3% on held-out business scenarios. These results show that scenario-grounded environments can provide useful and generalizable learning signals beyond benchmark-specific training, while their construction can itself become a learnable capability.
Minbyul Jeong, Chanwoong Yoon
Meridian Intelligence Global Inc. · University of Massachusetts Amherst
Executable environments are critical for post-training agents on tasks that require tool use and multi-step interaction, but constructing executable tasks together with their environments remains difficult to scale. Skills provide reusable domain knowledge, operational procedures, and tool-use instructions, but a substantial gap remains between the information contained in a skill and a concrete, challenging task with a complete executable environment. To address this gap, we introduce Skill2Env, a capability-oriented framework that starts from a skill and uses agent capability demands to guide task and environment synthesis. Skill2Env represents these demands through reusable difficulty patterns and instantiates them into task blueprints that specify objectives, challenges, environment facts, information boundaries, and acceptance criteria. These blueprints guide the joint construction of task instructions, execution substrates, workspaces, and rubric-based evaluators around source skills. We further propose Iterative Task Hardening, which uses solver execution evidence to identify insufficiently challenging task designs, strengthen or extend their difficulty-pattern instantiations, and revise the corresponding blueprints and environments. Using 1.5K high-scoring trajectories generated from Skill2Env environments for supervised fine-tuning, we observe consistent improvements across a broad range of agent benchmarks, demonstrating the effectiveness of capability-oriented environment synthesis for agent post-training.
The paradigm of LLMs has rapidly shifted from passive language interfaces to autonomous Claw-like agents that execute long-horizon tasks across stateful workspaces. While Agentic Reinforcement Learning (Agentic RL) provides a promising path to optimize these agents, its scaling is heavily bottlenecked by the severe scarcity of interactive training environments. Existing synthetic environments are strictly limited to tool-calling endpoints, rendering them insufficient for accommodating the end-to-end real-world demands of claw-like agents. To bridge this gap, we introduce EnvCraft, an automated framework for synthesizing executable environments and scalable training data. Specifically, EnvCraft employs an environment synthesis engine to build sandbox-isolated workspaces, alongside a topology-aware data generation engine to produce coherent task trajectories. Overall, we synthesize 139 interactive environments comprising approximately 20K complex tasks for Agentic RL training. Experiments on Qwen3/3.5 models (8B-32B) show that our method yields gains of up to +11.9% on Claw-style benchmarks and +8.0% on general tool-use benchmarks, with concurrent reductions in inference token cost. The results confirm that synthesized executable environments provide robust and generalizable learning signals for training.
Yirong Zeng, Shen You, Jinhang Feng +8
Harbin Institute of Technology, SCIR Lab · Huawei Technologies Co., Ltd · Peking University +1