Prompts

MCP prompts appear as slash commands in clients that support them (Claude Desktop, Cursor, and others). They bundle the right tool calls and analysis instructions for recurring workflows, so a full report is one command away.

Prompt	What it produces
`/weekly_pulse`	Week-over-week digest of your brand metrics: what moved and why
`/competitor_radar`	Competitor movement scan across brands, topics, and engines
`/engine_scorecard`	Per-engine breakdown: visibility, Share of Voice, sentiment, position, sources
`/topic_heatmap`	Visibility heatmap across topics × AI engines: where you’re strong and where weak
`/prompt_grader`	Grades your prompt set: topic balance, tag hygiene, funnel coverage, duplicates
`/source_authority`	Domain authority audit: retrieval rates, citation rates, top URLs
`/campaign_tracker`	Before/after comparison of your metrics around a campaign launch date

All prompts take project_id as the first argument; several accept optional date-range or brand filters. Your client shows each prompt’s argument list.

/weekly_pulse project_id=<uuid>

Prompts are read-only: they never modify your project, so they’re safe to run with just the mcp:read scope.

Get Started

Developers

Set Up Your Project

Interpret Your Results

Take Action

Integrations

MCP

Account & Billing