SigNoz Docs

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

Access Docs

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

Preferred: MCP tools

• signoz_search_docs: keyword search over the indexed docs corpus. Pass 2 to 6 keywords for one topic as searchText, for example Kubernetes pod logs; keep product and technology names as the user wrote them and drop filler such as "how do I" or "please". Do not pass the whole question sentence. When the question contains only one usable term, such as Retention?, send that term alone rather than padding or inventing keywords. Split a question with two intents into two calls. Narrow with section_slug when the question maps cleanly to a single docs section; reuse a section_slug from an earlier result rather than guessing one, and defer to the tool's parameter descriptions for slug and ranking behavior. If the top result is off-topic, retry once with fewer or different terms before answering; for a single-term question, retry with a synonym or one added contextual term, such as data retention for Retention?.
• signoz_fetch_doc: markdown for one indexed page. Pass the canonical URL or /docs/... path; optionally narrow to a section with heading. Inspect truncation_reason and available_headings in every response rather than assuming the returned content is complete.

Never call signoz_search_docs with empty searchText; pass keywords taken from the user's question, adding a synonym or contextual term only for the retry described above. Never call signoz_fetch_doc without a URL; neither tool guesses missing input. signoz://... URIs are MCP resources: read them through the MCP resource API, never signoz_fetch_doc, which accepts only https://signoz.io/docs/... URLs or /docs/... paths and rejects other scopes.