serp - SKILL.md Agent Skill

name: serp description: Use when the user asks "what's Google showing for X", "analyze the search results for Y", "what's the AI Overview saying about Z", "what are people also asking about W", or any SERP-flavored intelligence ask. Surfaces AI Overview, organic results, and People-Also-Asked separately rather than as free-form prose. Optional drill-down via web_fetch on specific results. argument-hint: [--country=`] [--language=]`


serp
SERP analysis using the Massive MCP web_search tool.
When to activate
The user wants SERP intelligence: "what's Google showing for", "analyze the search results for", "what's the AI Overview saying about", "what are people also asking about", "what's ranking for".
When invoked explicitly via /massive:serp, the query is provided as: $ARGUMENTS
Tool preference
Use web_search from the massive MCP server. The response includes the AI Overview, organic results, and People-Also-Asked block as separate fields.
Workflow

Parse flags. If $ARGUMENTS (or the prompt) contains --country=<code> or --language=<code>, capture them.
Detect geo/locale intent. Populate country, subdivision, city, language if the user named any (and no explicit flag overrides).
Search. Call web_search with the user's query plus geo params.
Surface as a structured digest. Output three sections:
AI Overview: the ai_overview text, with its source URLs.
Organic results: numbered list of {title} — {snippet} ({url}).
People also ask: numbered list of question + answer pairs.


Optional drill-down. If the user asks for depth on a specific result or the AI Overview source, web_fetch it.

Output as a structured digest, not free-form prose. Headings and lists matter.