# Grok Search

> Use Grok Search MCP for planned web research: define intent, assess complexity, break the task into sub-queries, and then search, fetch, or map sources. Use it for multi-source research that benefits from explicit planning, and state which Grok model is being used whenever that is known.

- Skill: `ry-run/grok-search` (Agent Skill, multi-file: 5 files)
- Install (CLI): `npx skillmds@latest add ry-run/grok-search`
- Raw SKILL.md: https://api.skillmd.com/api/skills/ry-run/grok-search/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: ry-run (https://skillmd.com/u/ry-run)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/ry-run/grok-search

---


# Grok Search

## Quick start

Goal: do deliberate web research instead of ad hoc searching when the question is broad, multi-part, or source-sensitive.

Reference notes: `skills/grok-search/references/grok-search.md`

## Required planning flow

Start with:
1. `mcp__grok_search__plan_intent`
2. `mcp__grok_search__plan_complexity`

Then continue based on complexity:
- Level 1: sub-queries only
- Level 2: add search terms and tool mapping
- Level 3: complete all phases including execution ordering

Execution tools:
- `mcp__grok_search__web_search`
- `mcp__grok_search__web_fetch`
- `mcp__grok_search__web_map`

## When to use it

- Research with multiple sub-questions.
- Tasks needing explicit source planning.
- Broad or ambiguous web investigations.
- Cases where site mapping or page fetching matters.

## Workflow

1. Plan the research.
2. Execute the right search/fetch/map actions.
3. If needed, inspect sources from a prior search with `get_sources`.

## Guardrails

- Do not skip `plan_intent`.
- State the Grok model explicitly when using this skill if the active model is known.
- Use `web_fetch` for full-page extraction, not search.
- Use `web_map` when you need site structure, not answer generation.

## References

- Condensed notes: `skills/grok-search/references/grok-search.md`