菜单
Assists developers in collecting and structuring a library of diverse examples ("Golden Dataset") required for data-driven evaluation, including tool trajectories.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Enforces data governance by verifying Knowledge Catalog Aspects before recommending BigQuery tables.
Ensures the game has a fun easter egg that turns the screen black and white, sepia or gameboy monotone colours. Use when creating or modifying a game.
Ensures platformer games have polished movement, responsive controls, and balanced jump mechanics. Use ONLY when the game type is "platformer".
Guides the creation of a new game and is the high-level orchestrator skill that MUST be invoked first. Use when tasked with creating a new game.
Arcade game must have a retro asthetic. Use when creating or modifying a game to ensure that the game looks retro and maintains a consistent look and feel with all games.
Game will always use the same control scheme/buttons and mapping of the gamepad and keyboard. Use when creating or modifying a game.
| name | gcp-agent-golden-dataset-builder |
| description | Assists developers in collecting and structuring a library of diverse examples ("Golden Dataset") required for data-driven evaluation, including tool trajectories. |
This skill helps you build the foundation for data-driven agent development: the Golden Dataset. Grounded in evaluation_blog.md, it focuses on verifying not just the final answer, but the "Thinking Process" (Reasoning Trace).
Ask Antigravity to:
A production-ready dataset uses the .jsonl format and includes:
prompt: The user input.reference: The ground truth answer (for semantic ResponseMatch).reference_trajectory: A list of expected tool calls. This allows the evaluator to check if the agent used the right tools in the right order.Refer to examples/trajectory_dataset.jsonl for the implementation. Note the use of tool_name and tool_input in the trajectory.