signoz-searching-docs

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

SigNoz Docs

Use official

signoz.io

documentation and API references only. Ground every answer in fetched docs content and cite the canonical docs URL.

仅使用官方

signoz.io

文档和API参考。所有回答都必须基于获取到的文档内容，并引用标准文档URL。

Access Docs

访问文档

Prefer the SigNoz MCP server tools when available; fall back to direct HTTP fetch.

优先使用可用的SigNoz MCP服务器工具；若不可用，则退回到直接HTTP请求。

Preferred: MCP tools

首选：MCP工具

```
signoz:signoz_search_docs
```
— BM25 search over the indexed docs corpus. Pass the user's natural-language query as
```
query
```
. Narrow with
```
section_slug
```
when the question maps cleanly to a single docs section (the tool's own schema lists valid slugs — defer to it rather than memorizing). Trust the ranking — the index handles relevance.
```
signoz:signoz_fetch_doc
```
— full markdown for one indexed page. Pass the canonical URL or
```
/docs/...
```
path; optionally narrow to a section with
```
heading
```
.

```
signoz:signoz_search_docs
```
— 对已索引的文档库进行BM25搜索。将用户的自然语言查询作为
```
query
```
传入。当问题明确对应单个文档章节时，使用
```
section_slug
```
缩小范围（工具自身的架构列出了有效的slug，请以其为准，无需记忆）。信任排序结果——索引会处理相关性。
```
signoz:signoz_fetch_doc
```
— 获取单个已索引页面的完整markdown内容。传入标准URL或
```
/docs/...
```
路径；可选择性地使用
```
heading
```
缩小到特定章节。

Fallback: direct HTTP fetch

备选：直接HTTP请求

If the MCP tools are unavailable, SigNoz docs support

Accept: text/markdown

natively.

Discover via the sitemap:

GET https://signoz.io/docs/sitemap.md

Fetch a specific page:

GET https://signoz.io/docs/<path>/
Accept: text/markdown

如果MCP工具不可用，SigNoz文档原生支持

Accept: text/markdown

请求头。

通过站点地图查找：

GET https://signoz.io/docs/sitemap.md

获取特定页面：

GET https://signoz.io/docs/<path>/
Accept: text/markdown

Workflow

工作流程

Identify the domain from the user's question: instrumentation, OpenTelemetry setup, querying, dashboards, alerts, troubleshooting, deployment, or API docs.
Check the heuristics table below. If a heuristic matches, read it before answering — heuristics encode product decisions (which path/method fits the user's environment), useful in both paths.
Search and fetch — pick the path based on tool availability:
- With MCP tools: call
```
signoz:signoz_search_docs
```
  with the user's query; pass
```
section_slug
```
  if the domain maps cleanly to one. Read the top 1-3 results and call
```
signoz:signoz_fetch_doc
```
  on the chosen URL (use
```
heading
```
  to narrow if the page is large and the question is sub-section-specific).
- Without MCP tools: grep
```
sitemap.md
```
  for candidate pages, rank the best 2-5 by how directly they answer the task, and
```
GET
```
  the top page(s) with
```
Accept: text/markdown
```
  . Heuristic coverage is sparse — for topics without a heuristic row, skim the sitemap by section path and prefer setup/troubleshooting/API-reference pages over overviews.
- Fetch one page for narrow questions; fetch multiple pages when the task spans setup + troubleshooting, or method-selection + language guide. Keep the set small.
Answer from the fetched docs and cite canonical
```
https://signoz.io/docs/...
```
URLs.
Handle ambiguity deliberately: if multiple pages are plausible, prefer the one that completes the task most directly; mention alternates only when they materially change the answer.

识别领域：从用户的问题中确定领域：instrumentation、OpenTelemetry设置、查询、仪表板、告警、故障排查、部署或API文档。
查看下方启发规则表：如果匹配某条启发规则，在回答前先阅读该规则——启发规则包含产品决策逻辑（哪种路径/方法适合用户环境），在两种访问路径中都有用。
搜索并获取文档：根据工具可用性选择路径：
- 有MCP工具时：调用
```
signoz:signoz_search_docs
```
  并传入用户的查询；如果领域明确对应某个章节，传入
```
section_slug
```
  。阅读前1-3条结果，并对选定的URL调用
```
signoz:signoz_fetch_doc
```
  （如果页面篇幅较长且问题针对子章节，可使用
```
heading
```
  缩小范围）。
- 无MCP工具时：在
```
sitemap.md
```
  中查找候选页面，根据与任务的匹配程度选出最佳的2-5个页面，然后通过
```
GET
```
  请求并携带
```
Accept: text/markdown
```
  头获取排名靠前的页面。启发规则的覆盖范围有限——对于没有对应启发规则行的主题，按章节路径浏览站点地图，优先选择设置/故障排查/API参考页面而非概述页面。
- 针对范围较窄的问题获取单个页面；当任务涉及设置+故障排查，或方法选择+语言指南时获取多个页面。页面数量请保持少量。
基于获取的文档回答，并引用标准的
```
https://signoz.io/docs/...
```
URL。
明确处理歧义：如果多个页面都看似合理，优先选择最直接完成任务的页面；仅当其他选项会实质性改变答案时才提及替代方案。

Message Actions

消息操作

On the terminal answer, emit FE-handoff actions per the SigNoz Skills & MCP spec:

open_docs
— include with the canonical URL of the primary cited page. Docs lookups are precisely the case where deep-linking to the source page helps the user read in context and verify the answer.
follow_up
— 1-2 next-step prompts that build on a docs answer. After a setup guide: "walk me through the first command" or "what's a common gotcha here?". After a concept page: "show me a worked example."
Do NOT emit
apply_filter
. Docs answers do not produce a query for an explorer page; emitting
```
apply_filter
```
would overwrite the user's working query.

Verbatim guardrail: When answering a SigNoz docs question, include an
open_docs
action on the final message with the canonical URL of the primary cited page.

在终端回答中，按照SigNoz技能与MCP规范触发FE-handoff操作：

open_docs
— 包含主要引用页面的标准URL。文档查询场景下，直接链接到源页面有助于用户在上下文环境中阅读并验证答案。
follow_up
— 1-2个基于文档答案的后续步骤提示。例如，在设置指南之后：“引导我完成第一条命令” 或 “这里常见的陷阱是什么？”。在概念页面之后：“给我展示一个实际示例。”
请勿触发
apply_filter
。文档回答不会生成用于探索页面的查询；触发
```
apply_filter
```
会覆盖用户当前的工作查询。

明确规则：回答SigNoz文档相关问题时，在最终消息中包含
open_docs
操作，并附上主要引用页面的标准URL。

Domain Heuristics

领域启发规则

Read the matching heuristic file before fetching docs. Each file contains decision logic to route the user to the right guide.

Topic	Trigger keywords	Heuristic file
Sending Logs	logs, log collection, logging, send logs	sending-logs.md

在获取文档前，请先阅读匹配的启发规则文件。每个文件包含决策逻辑，用于引导用户找到正确的指南。

主题	触发关键词	启发规则文件
发送日志	logs、日志收集、logging、send logs	sending-logs.md