npx skills add https://github.com/arize-ai/arize-skills --skill arize-link生成指向 Arize UI 中追踪、跨度和会话的深层链接。
从用户或上下文(例如,导出的追踪数据、解析的 URL)中收集以下信息:
基础 URL:https://app.arize.com(如果用户指定了自定义基础 URL,则覆盖以用于本地部署)
打开追踪侧滑面板,显示追踪中的所有跨度。
{base_url}/organizations/{org_id}/spaces/{space_id}/projects/{project_id}?selectedTraceId={trace_id}&queryFilterA=&selectedTab=llmTracing&timeZoneA=America%2FLos_Angeles&startA={start_epoch_ms}&endA={end_epoch_ms}&envA=tracing&modelType=generative_llm
广告位招租
在这里展示您的产品或服务
触达数万 AI 开发者,精准高效
如果同时提供了 span_id,则添加 &selectedSpanId={span_id} 以在追踪中高亮显示该跨度。
在追踪中打开特定的跨度。需要同时提供 trace_id 和 span_id。
{base_url}/organizations/{org_id}/spaces/{space_id}/projects/{project_id}?selectedTraceId={trace_id}&selectedSpanId={span_id}&queryFilterA=&selectedTab=llmTracing&timeZoneA=America%2FLos_Angeles&startA={start_epoch_ms}&endA={end_epoch_ms}&envA=tracing&modelType=generative_llm
打开对话/交互流程的会话视图。
{base_url}/organizations/{org_id}/spaces/{space_id}/projects/{project_id}?selectedSessionId={session_id}&queryFilterA=&selectedTab=llmTracing&timeZoneA=America%2FLos_Angeles&startA={start_epoch_ms}&endA={end_epoch_ms}&envA=tracing&modelType=generative_llm
重要:startA 和 endA 是必需的查询参数。没有它们,Arize UI 将默认显示最近 7 天,如果追踪/跨度不在该时间窗口内,则会显示“您的模型没有任何近期数据”错误。
按以下优先级顺序使用这些来源:
用户提供的 URL:如果用户分享了 Arize URL,则从中提取 startA 和 endA 并复用。这是最可靠的方法,因为它保留了用户原始的时间窗口。
导出的跨度数据:如果您有跨度数据(例如,来自 ax spans export),则使用跨度的 start_time 字段来计算一个覆盖该数据的时间范围:
python -c " from datetime import datetime, timedelta t = datetime.fromisoformat('2026-03-07T05:39:15.822147Z'.replace('Z','+00:00')) start = int((t - timedelta(days=1)).timestamp() * 1000) end = int((t + timedelta(days=1)).timestamp() * 1000) print(f'startA={start}&endA={end}') "
默认回退方案:使用最近 90 天。计算如下:
* `startA`: `(当前时间 - 90 天)` 转换为纪元毫秒
* `endA`: 当前时间转换为纪元毫秒
startA 和 endA 的纪元毫秒值。给定:org_id=QWNjb3VudE9yZ2FuaXphdGlvbjoxOmFiQzE=, space_id=U3BhY2U6MTp4eVo5, project_id=TW9kZWw6MTpkZUZn, trace_id=0123456789abcdef0123456789abcdef
追踪链接:
https://app.arize.com/organizations/QWNjb3VudE9yZ2FuaXphdGlvbjoxOmFiQzE=/spaces/U3BhY2U6MTp4eVo5/projects/TW9kZWw6MTpkZUZn?selectedTraceId=0123456789abcdef0123456789abcdef&queryFilterA=&selectedTab=llmTracing&timeZoneA=America%2FLos_Angeles&startA=1700000000000&endA=1700086400000&envA=tracing&modelType=generative_llm
每周安装量
18
代码仓库
GitHub 星标数
2
首次出现
1 天前
安全审计
安装于
opencode18
gemini-cli18
github-copilot18
amp18
cline18
codex18
Generate deep links to the Arize UI for traces, spans, and sessions.
Collect these from the user or from context (e.g., exported trace data, parsed URLs):
Base URL: https://app.arize.com (override for on-prem if the user specifies a custom base URL)
Opens the trace slideover showing all spans in the trace.
{base_url}/organizations/{org_id}/spaces/{space_id}/projects/{project_id}?selectedTraceId={trace_id}&queryFilterA=&selectedTab=llmTracing&timeZoneA=America%2FLos_Angeles&startA={start_epoch_ms}&endA={end_epoch_ms}&envA=tracing&modelType=generative_llm
If a span_id is also available, add &selectedSpanId={span_id} to highlight that span within the trace.
Opens a specific span within a trace. Both trace_id and span_id are required.
{base_url}/organizations/{org_id}/spaces/{space_id}/projects/{project_id}?selectedTraceId={trace_id}&selectedSpanId={span_id}&queryFilterA=&selectedTab=llmTracing&timeZoneA=America%2FLos_Angeles&startA={start_epoch_ms}&endA={end_epoch_ms}&envA=tracing&modelType=generative_llm
Opens the session view for a conversation/interaction flow.
{base_url}/organizations/{org_id}/spaces/{space_id}/projects/{project_id}?selectedSessionId={session_id}&queryFilterA=&selectedTab=llmTracing&timeZoneA=America%2FLos_Angeles&startA={start_epoch_ms}&endA={end_epoch_ms}&envA=tracing&modelType=generative_llm
CRITICAL : startA and endA are required query parameters. Without them, the Arize UI defaults to the last 7 days and will show a "Your model doesn't have any recent data" error if the trace/span falls outside that window.
Use these sources in priority order:
User-provided URL : If the user shared an Arize URL, extract startA and endA from it and reuse them. This is the most reliable approach since it preserves the user's original time window.
Exported span data : If you have span data (e.g., from ax spans export), use the span's start_time field to calculate a range that covers the data:
# Convert span start_time to epoch ms, then pad ±1 day
python -c "
from datetime import datetime, timedelta
t = datetime.fromisoformat('2026-03-07T05:39:15.822147Z'.replace('Z','+00:00'))
start = int((t - timedelta(days=1)).timestamp() * 1000)
end = int((t + timedelta(days=1)).timestamp() * 1000)
print(f'startA={start}&endA={end}')
"
Default fallback : Use the last 90 days. Calculate:
startA: as epoch millisecondsstartA and endA epoch milliseconds using the priority order above.Given: org_id=QWNjb3VudE9yZ2FuaXphdGlvbjoxOmFiQzE=, space_id=U3BhY2U6MTp4eVo5, project_id=TW9kZWw6MTpkZUZn, trace_id=0123456789abcdef0123456789abcdef
Trace link:
https://app.arize.com/organizations/QWNjb3VudE9yZ2FuaXphdGlvbjoxOmFiQzE=/spaces/U3BhY2U6MTp4eVo5/projects/TW9kZWw6MTpkZUZn?selectedTraceId=0123456789abcdef0123456789abcdef&queryFilterA=&selectedTab=llmTracing&timeZoneA=America%2FLos_Angeles&startA=1700000000000&endA=1700086400000&envA=tracing&modelType=generative_llm
Weekly Installs
18
Repository
GitHub Stars
2
First Seen
1 day ago
Security Audits
Gen Agent Trust HubPassSocketPassSnykPass
Installed on
opencode18
gemini-cli18
github-copilot18
amp18
cline18
codex18
AI 代码实施计划编写技能 | 自动化开发任务分解与 TDD 流程规划工具
50,900 周安装
SEO结构架构专家:优化网站信息架构、标题层次与结构化数据
201 周安装
Elastic Observability SLO管理指南:创建、监控服务等级目标与SLI类型详解
204 周安装
CSS开发指南:Flexbox、Grid布局、响应式设计与性能优化最佳实践
199 周安装
OpenSpec (OPSX) 指南:基于工件的开发工作流系统,实现变更管理与自动化
202 周安装
React Testing Library 测试最佳实践:用户行为测试、查询优先级与异步处理指南
200 周安装
Web无障碍测试指南:WCAG合规、键盘导航与屏幕阅读器测试最佳实践
202 周安装
(now - 90 days)endA: current time as epoch milliseconds