智能体
skills.sh ·
Google agents-cli 评估技能
Google agents-cli 的 eval 技能,用于评估,安装量 9,682。
AI推荐理由 评估技能对 Agent 开发有价值,但摘要未提供具体方法。
60
安装 354
榜单名次 2
正文
google-agents-cli-eval
安装
命令提示符
$ npx skills add https://github.com/google/agents-cli --skill google-agents-cli-eval
摘要
使用指标、评估集和迭代式评估-修复循环来评估ADK代理。
使用agents-cli eval run运行评估,可配置标准(工具轨迹、响应匹配、基于评分标准的评分、幻觉检测、安全检查)和匹配类型(EXACT、IN_ORDER、ANY_ORDER)
构建包含多轮对话案例、预期工具轨迹、中间响应和会话状态覆盖的评估集
迭代5-10+个评估-修复周期:诊断失败、修复代理指令或工具逻辑、重新运行,并使用任务列表跟踪进度
避免常见陷阱:不要降低阈值来隐藏失败,使用IN_ORDER匹配处理额外的工具调用,确保应用名称与目录匹配,并使用回调初始化状态以防止KeyError崩溃
SKILL.md
代理评估指南
要求:agents-cli(uv tool install google-agents-cli)——如果需要,先安装uv。
脚手架项目?如果你使用了/google-agents-cli-scaffold,你已经拥有agents-cli eval run(链式生成+评分)、tests/eval/datasets/和tests/eval/eval_config.yaml。从执行eval run开始,并在此基础上迭代。
参考文件
文件
内容
references/dataset_schema.md
规范EvaluationDataset模式——所有字段类型、单轮/多轮/多代理的JSON示例、常见错误
references/metrics-guide.md
完整指标参考——所有内置指标、匹配类型、自定义指标、评判模型配置
references/user-simulation.md
动态对话测试——评估数据集合成标志、场景是什么、兼容指标
references/builtin-tools-eval.md
google_search和模型内部工具——轨迹行为、指标兼容性
references/advanced-commands.md
可选命令:eval analyze、eval optimize、eval submit / eval results
references/multimodal-eval.md
多模态输入——评估数据集模式、内置指标限制、自定义评估器模式
显示更多
安装量
155.4K
仓库
google/agents-cli
GitHub星标
5.8K
首次出现
2026年4月21日
安全审计
Gen Agent Trust Hub通过
Socket通过
Snyk警告
查看原文(中文正文为 AI 翻译)
google-agents-cli-eval
Installation
CommandPrompt
$ npx skills add https://github.com/google/agents-cli --skill google-agents-cli-eval
Summary
Evaluate ADK agents with metrics, evalsets, and the iterative eval-fix loop.
Run evaluations with agents-cli eval run using configurable criteria (tool trajectory, response matching, rubric-based scoring, hallucination detection, safety checks) and match types (EXACT, IN_ORDER, ANY_ORDER)
Build evalsets with multi-turn conversation cases, expected tool trajectories, intermediate responses, and session state overrides
Iterate through 5-10+ eval-fix cycles: diagnose failures, fix agent instructions or tool logic, rerun, and track progress with task lists
Avoid common pitfalls: don't lower thresholds to hide failures, handle extra tool calls with IN_ORDER matching, ensure app name matches directory, and initialize state with callbacks to prevent KeyError crashes
SKILL.md
Agent Evaluation Guide
Requires: agents-cli (uv tool install google-agents-cli) — install uv first if needed.
Scaffolded project? If you used /google-agents-cli-scaffold, you already have agents-cli eval run (chains generate + grade), tests/eval/datasets/, and tests/eval/eval_config.yaml. Start with executing eval run and iterate from there.
Reference Files
File
Contents
references/dataset_schema.md
Canonical EvaluationDataset schema — all field types, JSON examples for single-turn / multi-turn / multi-agent, common mistakes
references/metrics-guide.md
Complete metrics reference — all built-in metrics, match types, custom metrics, judge model config
references/user-simulation.md
Dynamic conversation testing — eval dataset synthesize flags, what scenarios are, compatible metrics
references/builtin-tools-eval.md
google_search and model-internal tools — trajectory behavior, metric compatibility
references/advanced-commands.md
Opt-in commands: eval analyze, eval optimize, eval submit / eval results
references/multimodal-eval.md
Multimodal inputs — eval dataset schema, built-in metric limitations, custom evaluator pattern
Show more
Installs
155.4K
Repository
google/agents-cli
GitHub Stars
5.8K
First Seen
Apr 21, 2026
Security Audits
Gen Agent Trust HubPass
SocketPass
SnykWarn