| name | dd-lookup-docs |
| description | Look up current Datadog documentation via the official LLM-optimized index at docs.datadoghq.com/llms.txt. Use this skill whenever you need accurate, up-to-date Datadog configuration steps, feature details, API references, integration setup, or troubleshooting guidance — even if you think you already know the answer. Datadog docs change frequently and your training data is likely stale. Trigger on any Datadog-related question: agent install, log pipelines, APM setup, DBM, monitors, dashboards, SLOs, integrations, RBAC, cloud cost, synthetics, LLM observability, etc. Also trigger when another dd- prefixed skill needs to reference official documentation.
|
| tools | ["web_fetch"] |
Datadog Documentation Lookup
Look up current Datadog documentation using the official LLM-optimized index.
When to Use
- Any time you need Datadog-specific configuration, setup, or feature details.
- When another dd- skill references this one for version-sensitive documentation.
- Always prefer this over training data — Datadog docs change frequently.
Procedure
Quick path — guess the URL first
Datadog docs follow a predictable URL structure. If you can guess the path, skip the index and fetch directly:
https://docs.datadoghq.com/<product>/<subtopic>.md
Common patterns:
| Topic | URL pattern |
|---|
| Product overview | /<product>.md → logs.md, tracing.md, database_monitoring.md |
| Sub-feature | /<product>/<feature>.md → logs/log_configuration/pipelines.md |
| Setup guide | /<product>/setup_<engine>/<variant>.md → database_monitoring/setup_mysql/rds.md |
| Integration | /integrations/<name>.md → integrations/amazon_web_services.md |
| API endpoint | /api/latest/<resource>.md → api/latest/logs-pipelines.md |
If the guessed URL returns a valid page, you're done. If it 404s or doesn't match, fall through to Step 1.
Step 1 — Fetch the documentation index
Fetch the plain-text index:
https://docs.datadoghq.com/llms.txt
This file has two sections:
- Introductory overview — a short description of Datadog and its major product areas, with links.
# Docs Pages listing — one entry per line in the format:
- [Page Title](https://docs.datadoghq.com/<path>.md): Short description
Important: The URLs in the index already end in .md. You can fetch them directly — do not append .md again.
Step 2 — Find relevant entries
Search the index for keywords related to your topic. Use multiple keyword variants to cast a wide net. For example, to find Database Monitoring docs, search for: database_monitoring, dbm, postgres, mysql, setup_postgres.
Each matching line gives you:
- A human-readable title
- A direct
.md URL you can fetch
- A brief description of the page content
Pick the entries whose title and description best match your question. If multiple entries look relevant (e.g. a product overview page plus a specific setup page), fetch the most specific one first.
Step 3 — Fetch the documentation page
Fetch the .md URL from the index entry. The response is clean Markdown suitable for reading directly.
Example — if the index contains:
- [Database Monitoring](https://docs.datadoghq.com/database_monitoring.md): Learn about Database Monitoring and get started
Fetch https://docs.datadoghq.com/database_monitoring.md.
Following links within a doc page
Documentation pages often link to sub-pages using regular (non-.md) URLs, and those URLs frequently have a trailing slash:
https://docs.datadoghq.com/logs/log_configuration/
https://docs.datadoghq.com/database_monitoring/setup_postgres/selfhosted
To convert these to Markdown URLs, apply two steps in order:
- Strip any trailing slash —
log_configuration/ → log_configuration
- Append
.md — log_configuration → log_configuration.md
Correct result:
https://docs.datadoghq.com/logs/log_configuration.md
https://docs.datadoghq.com/database_monitoring/setup_postgres/selfhosted.md
Warning: Skipping step 1 produces log_configuration/.md, which will fail. Always strip the trailing slash first.
Detecting a failed fetch
If a guessed or constructed URL is wrong, you'll typically get:
- An HTTP 404 error
- An HTML page instead of Markdown (look for
<!DOCTYPE or <html> at the start)
- A redirect to the Datadog docs search page
If any of these happen, fall back to searching the llms.txt index or use web search as described in Step 4.
Step 4 — Fallback
If llms.txt is unreachable or the topic doesn't appear in the index:
- Try constructing a URL directly:
https://docs.datadoghq.com/<topic>.md
(e.g. https://docs.datadoghq.com/tracing.md)
- For API-specific questions, try:
https://docs.datadoghq.com/api/latest.md
Tips
- The index is very large (~200+ entries). It will often be truncated by your HTTP tool. If your topic keywords aren't found in the visible portion, try fetching with a higher token limit, or use the Quick Path approach to guess the URL directly.
- Expect 2–3 hops for specific topics. Overview pages rarely have the details you need — they link to sub-pages that do. Budget for following 1–2 in-page links.
- Prefer the most specific page. A setup guide for "MySQL on RDS" is far more useful than the generic "Database Monitoring" overview.
- In-page links often have trailing slashes — you must strip the slash before appending
.md. See "Following links within a doc page" for the exact two-step rule. Index URLs already end in .md — don't double-append.
- The
integrations.md page is a JSON catalog, not a readable listing. Don't fetch it to browse integrations. Instead, guess the URL directly: integrations/<name>.md (e.g. integrations/postgres.md, integrations/amazon_web_services.md). Integration names use lowercase with underscores or hyphens.
- The Markdown output uses custom directives like
{% callout %}, {% tab %}, {% image %}, and {% alert %}. These are readable enough to extract information from; don't be confused by them.
- If your HTTP tool restricts URLs (e.g. only allows previously-seen URLs), use a web search for
site:docs.datadoghq.com <topic> as a fallback to discover the right page URL, then fetch its .md variant.