Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2605.07442 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910200829575168 |
|---|---|
| author | Jia, Chaobo Wan, Ruipeng Sun, Ting Tan, Weihao Wan, Borui Tong, Yuxuan Sheng, Guangming Xu, Hong |
| author_facet | Jia, Chaobo Wan, Ruipeng Sun, Ting Tan, Weihao Wan, Borui Tong, Yuxuan Sheng, Guangming Xu, Hong |
| contents | LLM-based game generation promises to turn natural-language specifications into executable games, but progress is limited by the lack of reliable automated verification. Unlike conventional code generation, game correctness is defined over long-horizon interaction: a game may appear correct while violating core mechanics such as state updates, interaction rules, and phase transitions. Existing Agent-as-a-Verifier approaches collapse verification into open-ended gameplay, making verdicts reachability-bound, time-consuming, coverage-limited, and sensitive to the agent's gameplay ability.
We present GameGen-Verifier, an automated verification paradigm for LLM-generated games that decomposes a specification into verifiable keypoints and grounds them into independent verification units. Each unit patches the game runtime into a concrete target state, executes a bounded interaction, and judges the outcome against the keypoint assertion. We implement GGV-Harness, a scalable agentic harness providing concurrency management, runtime isolation, and fault recovery.
On VeriGame, our dataset of 100 games across seven genres, GameGen-Verifier achieves up to 92.2% accuracy against human judgments versus 58.8% for the coverage-enforced Agent-as-a-Verifier baseline, while reducing wall-clock time by up to 16.6x. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_07442 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | GameGen-Verifier: Parallel Keypoint-Based Verification for LLM-Generated Games via Runtime State Injection Jia, Chaobo Wan, Ruipeng Sun, Ting Tan, Weihao Wan, Borui Tong, Yuxuan Sheng, Guangming Xu, Hong Machine Learning LLM-based game generation promises to turn natural-language specifications into executable games, but progress is limited by the lack of reliable automated verification. Unlike conventional code generation, game correctness is defined over long-horizon interaction: a game may appear correct while violating core mechanics such as state updates, interaction rules, and phase transitions. Existing Agent-as-a-Verifier approaches collapse verification into open-ended gameplay, making verdicts reachability-bound, time-consuming, coverage-limited, and sensitive to the agent's gameplay ability. We present GameGen-Verifier, an automated verification paradigm for LLM-generated games that decomposes a specification into verifiable keypoints and grounds them into independent verification units. Each unit patches the game runtime into a concrete target state, executes a bounded interaction, and judges the outcome against the keypoint assertion. We implement GGV-Harness, a scalable agentic harness providing concurrency management, runtime isolation, and fault recovery. On VeriGame, our dataset of 100 games across seven genres, GameGen-Verifier achieves up to 92.2% accuracy against human judgments versus 58.8% for the coverage-enforced Agent-as-a-Verifier baseline, while reducing wall-clock time by up to 16.6x. |
| title | GameGen-Verifier: Parallel Keypoint-Based Verification for LLM-Generated Games via Runtime State Injection |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2605.07442 |