Reference
IQRush MCP: available tools and capabilities
All 43 tools across 8 families — what each does and when to use it.
IQRush MCP exposes 43 read-only tools across 8 families. All tools return live data scoped to your authenticated workspace. Every metric is returned with a 95% confidence interval — always show the full value, CI bounds, and method when presenting results.
Where to start
Always call start_session first. It bootstraps your workspace, returns your project list, and — once you specify a project — auto-fires a summary report covering headline KPIs and significant findings.
After start_session, the most useful next calls for a first analysis are:
• get_citation_score — your citation share by platform
• get_visibility_alignment — a composite brand health diagnostic in one call
• get_significant_findings — what changed, without specifying a metric
Tool families
Family 1 — Session & Discovery (8 tools)
Tool | What it does |
|---|---|
| Always the first tool called. Returns project list or auto-fires the summary report when a project is selected. |
| Lists all projects in the workspace. |
| Project metadata and brand context — owned domains, entity bundle, competitor bundles. |
| Measurement runs for a project, newest first. Always use the latest run where |
| Full run detail — status, data readiness, and counts. |
| Saved baseline snapshots for drift comparison. |
| Valid |
| Valid query-per-topic budgets for statistically balanced study design. |
Family 2 — Core Measurement (5 tools)
Tool | What it does |
|---|---|
| Citation share or coverage per domain/URL with confidence intervals. Always run with |
| Brand mention coverage per entity or bundle with confidence intervals. Use |
| Positive/negative/neutral rates and net sentiment per entity with confidence intervals. |
| Fraction of brand mentions backed by a citation. High = cited when named; low = named without a source. |
| Composite brand health diagnostic. Call this first for brand health questions — do not compose mention + groundedness manually. |
Family 3 — Grounding & Sources (5 tools)
Tool | What it does |
|---|---|
| Sources grounding brand mentions, ranked by share. |
| Which brands a given domain grounds, with sentiment distribution. |
| Shared vs. self-specific vs. competitor-specific grounding sources. Competitor-specific = content gap; self-specific = your moat. |
| The actual text of statements where brands are mentioned. Qualitative evidence only — no confidence intervals. |
| Query catalog with prompt type and intent labels. |
Family 4 — Analytical Synthesis (7 tools)
Bootstrap-backed. Expect 2–10s response times. All require at least 2 runs except get_results_stability.
Tool | What it does |
|---|---|
| Intra-run convergence scoring. Check |
| Fast drift screen between two runs. |
| Per-domain citation-share differences with confidence intervals. Use after |
| Statistical contrast for one domain or entity. Always check |
| Per-domain rank change with confidence interval. Prefer |
| Decompose one domain’s performance by platform, topic, intent, or prompt type. |
| Scan for citation-share changes and rank displacements across the run. Use when asked “what changed?” |
Family 5 — Timeseries (3 tools)
Requires at least 2 runs.
Tool | What it does |
|---|---|
| Track one metric across up to 24 runs with 95% confidence interval per point. |
| First vs. latest value, delta, direction, and volatility per series. |
| Top risers and fallers by first-to-latest delta. |
Family 6 — Technical (3 tools)
Tool | What it does |
|---|---|
| Raw pipeline feature scores for a URL: Core Web Vitals, page speed, structured data, indexability, and more. |
| Actionable fix recommendations, sorted by impact then effort. |
| Content-alignment analysis: identifies hotspots (sections that support AI claims) and weakspots (claims without content support). |
Family 7 — Convenience (3 tools)
Tool | What it does |
|---|---|
| Recipe dispatcher. Returns a structured analytics plan. Only call when the user explicitly asks to “start analytics mode” or “run the full report.” |
| Headline citation share, citation coverage, and mention count for a completed run. Precomputed. |
| Top URLs by citation share with confidence intervals. Precomputed. |
Family 8 — Google Analytics 4 (9 tools)
These tools return data only if a GA4 property is connected in IQRush Settings → Integrations → GA4.
Tool | What it does |
|---|---|
| All GA4 properties linked to the connected Google account. |
| Web traffic report — sessions, users, bounce rate, page views, conversions. Flexible date range. |
| Live active users and real-time sessions by country and device. |
| Available dimensions and metrics. Call before get_ga4_report to check what’s available. |
| AI traffic performance by source (ChatGPT, Perplexity, Gemini, Claude, Copilot). Primary tool for AI traffic questions. |
| E-commerce conversion funnel driven by AI traffic: view → cart → checkout → purchase. |
| Product-level revenue and conversion data for AI-referred sessions. |
| Reference list of AI platforms tracked by IQRush: name, category, GA4 source patterns. |
| Google Search Console query performance — clicks, impressions, CTR, position. Nightly sync. |
Analytics mode bundles
The analytics_mode tool accepts a bundle parameter. Only call it when the user explicitly asks to “start analytics mode” or “run the full report.”
Bundle | What it covers |
|---|---|
| Headline KPIs and significant findings (default; also the login report) |
| Full citation landscape: share, coverage, top domains/URLs, conquest sources |
| Presence and prominence: count, coverage, groundedness, entity network |
| Tone: net + per-class with confidence intervals, by platform, verbatim evidence |
| Prompt catalog and per-prompt brand performance gaps |
| Topic prioritisation quadrant with confidence interval whiskers, factor analysis |
| Cross-engine matrix, per-platform charts, paired significance tests |
| Run convergence verdict, yield by platform, sample-size sufficiency |
| What changed vs. a baseline run (requires baseline_id) |
| Methodology panel and run integrity audit |
| Content and technical action items |
| Citation map for owned pages |
| All 12 bundles in sequence |
Frequently asked questions
Do I need to call start_session every time?
Yes. start_session bootstraps your workspace and resolves the latest run. It is always the first tool called — the connector handles this automatically when you start a conversation.
Can IQRush MCP modify my projects or data?
No. All 43 tools are read-only. The connector cannot create, modify, or delete projects, runs, settings, or brand context.
What happens if my token expires?
Device flow tokens (Claude Desktop) expire after 90 days. OAuth tokens (remote clients) are managed by the client. If you’re prompted to sign in again, your token has expired. For Claude Desktop, restart the app and follow the sign-in link again.
Does IQRush MCP work with AI tools not on the confirmed list?
Any MCP-compatible client that supports HTTP transport with OAuth 2.1 can connect using the server URL. Add the URL directly in your client’s MCP server settings. Check your client’s documentation for where to add remote MCP servers.
Why do some tools take longer to respond?
The analytical synthesis tools (Family 4) run bootstrap-backed statistical computations. Expect 2–10 seconds. Other tools return precomputed results and respond faster.
IQRush MCP exposes 43 read-only tools across 8 families. All tools return live data scoped to your authenticated workspace. Every metric is returned with a 95% confidence interval — always show the full value, CI bounds, and method when presenting results.
Where to start
Always call start_session first. It bootstraps your workspace, returns your project list, and — once you specify a project — auto-fires a summary report covering headline KPIs and significant findings.
After start_session, the most useful next calls for a first analysis are:
• get_citation_score — your citation share by platform
• get_visibility_alignment — a composite brand health diagnostic in one call
• get_significant_findings — what changed, without specifying a metric
Tool families
Family 1 — Session & Discovery (8 tools)
Tool | What it does |
|---|---|
| Always the first tool called. Returns project list or auto-fires the summary report when a project is selected. |
| Lists all projects in the workspace. |
| Project metadata and brand context — owned domains, entity bundle, competitor bundles. |
| Measurement runs for a project, newest first. Always use the latest run where |
| Full run detail — status, data readiness, and counts. |
| Saved baseline snapshots for drift comparison. |
| Valid |
| Valid query-per-topic budgets for statistically balanced study design. |
Family 2 — Core Measurement (5 tools)
Tool | What it does |
|---|---|
| Citation share or coverage per domain/URL with confidence intervals. Always run with |
| Brand mention coverage per entity or bundle with confidence intervals. Use |
| Positive/negative/neutral rates and net sentiment per entity with confidence intervals. |
| Fraction of brand mentions backed by a citation. High = cited when named; low = named without a source. |
| Composite brand health diagnostic. Call this first for brand health questions — do not compose mention + groundedness manually. |
Family 3 — Grounding & Sources (5 tools)
Tool | What it does |
|---|---|
| Sources grounding brand mentions, ranked by share. |
| Which brands a given domain grounds, with sentiment distribution. |
| Shared vs. self-specific vs. competitor-specific grounding sources. Competitor-specific = content gap; self-specific = your moat. |
| The actual text of statements where brands are mentioned. Qualitative evidence only — no confidence intervals. |
| Query catalog with prompt type and intent labels. |
Family 4 — Analytical Synthesis (7 tools)
Bootstrap-backed. Expect 2–10s response times. All require at least 2 runs except get_results_stability.
Tool | What it does |
|---|---|
| Intra-run convergence scoring. Check |
| Fast drift screen between two runs. |
| Per-domain citation-share differences with confidence intervals. Use after |
| Statistical contrast for one domain or entity. Always check |
| Per-domain rank change with confidence interval. Prefer |
| Decompose one domain’s performance by platform, topic, intent, or prompt type. |
| Scan for citation-share changes and rank displacements across the run. Use when asked “what changed?” |
Family 5 — Timeseries (3 tools)
Requires at least 2 runs.
Tool | What it does |
|---|---|
| Track one metric across up to 24 runs with 95% confidence interval per point. |
| First vs. latest value, delta, direction, and volatility per series. |
| Top risers and fallers by first-to-latest delta. |
Family 6 — Technical (3 tools)
Tool | What it does |
|---|---|
| Raw pipeline feature scores for a URL: Core Web Vitals, page speed, structured data, indexability, and more. |
| Actionable fix recommendations, sorted by impact then effort. |
| Content-alignment analysis: identifies hotspots (sections that support AI claims) and weakspots (claims without content support). |
Family 7 — Convenience (3 tools)
Tool | What it does |
|---|---|
| Recipe dispatcher. Returns a structured analytics plan. Only call when the user explicitly asks to “start analytics mode” or “run the full report.” |
| Headline citation share, citation coverage, and mention count for a completed run. Precomputed. |
| Top URLs by citation share with confidence intervals. Precomputed. |
Family 8 — Google Analytics 4 (9 tools)
These tools return data only if a GA4 property is connected in IQRush Settings → Integrations → GA4.
Tool | What it does |
|---|---|
| All GA4 properties linked to the connected Google account. |
| Web traffic report — sessions, users, bounce rate, page views, conversions. Flexible date range. |
| Live active users and real-time sessions by country and device. |
| Available dimensions and metrics. Call before get_ga4_report to check what’s available. |
| AI traffic performance by source (ChatGPT, Perplexity, Gemini, Claude, Copilot). Primary tool for AI traffic questions. |
| E-commerce conversion funnel driven by AI traffic: view → cart → checkout → purchase. |
| Product-level revenue and conversion data for AI-referred sessions. |
| Reference list of AI platforms tracked by IQRush: name, category, GA4 source patterns. |
| Google Search Console query performance — clicks, impressions, CTR, position. Nightly sync. |
Analytics mode bundles
The analytics_mode tool accepts a bundle parameter. Only call it when the user explicitly asks to “start analytics mode” or “run the full report.”
Bundle | What it covers |
|---|---|
| Headline KPIs and significant findings (default; also the login report) |
| Full citation landscape: share, coverage, top domains/URLs, conquest sources |
| Presence and prominence: count, coverage, groundedness, entity network |
| Tone: net + per-class with confidence intervals, by platform, verbatim evidence |
| Prompt catalog and per-prompt brand performance gaps |
| Topic prioritisation quadrant with confidence interval whiskers, factor analysis |
| Cross-engine matrix, per-platform charts, paired significance tests |
| Run convergence verdict, yield by platform, sample-size sufficiency |
| What changed vs. a baseline run (requires baseline_id) |
| Methodology panel and run integrity audit |
| Content and technical action items |
| Citation map for owned pages |
| All 12 bundles in sequence |
Frequently asked questions
Do I need to call start_session every time?
Yes. start_session bootstraps your workspace and resolves the latest run. It is always the first tool called — the connector handles this automatically when you start a conversation.
Can IQRush MCP modify my projects or data?
No. All 43 tools are read-only. The connector cannot create, modify, or delete projects, runs, settings, or brand context.
What happens if my token expires?
Device flow tokens (Claude Desktop) expire after 90 days. OAuth tokens (remote clients) are managed by the client. If you’re prompted to sign in again, your token has expired. For Claude Desktop, restart the app and follow the sign-in link again.
Does IQRush MCP work with AI tools not on the confirmed list?
Any MCP-compatible client that supports HTTP transport with OAuth 2.1 can connect using the server URL. Add the URL directly in your client’s MCP server settings. Check your client’s documentation for where to add remote MCP servers.
Why do some tools take longer to respond?
The analytical synthesis tools (Family 4) run bootstrap-backed statistical computations. Expect 2–10 seconds. Other tools return precomputed results and respond faster.
© 2026 IQRush. All Rights Reserved.
Site by ONBOX
AI search visibility you can defend
Whether you're building, buying, or briefing on AI search, get decision-grade data that holds.
Resources
AI search visibility you can defend
Whether you're building, buying, or briefing on AI search, get decision-grade data that holds.
© 2026 IQRush. All Rights Reserved.