Integration · Semrush

Keyword research, position tracking, and SEO-PR through one MCP.

The widely-used SEO platform with stronger US keyword data than Ahrefs and SEO-PR features Ahrefs does not cover. The MCP exposes keyword research, position tracking, traffic analytics, backlink analytics, and content optimization tools.

Topic Research for content brief authoring, Position Tracking with daily SERP-position and SERP-feature deltas, Keyword Magic Tool for cluster expansion, Brand Monitoring for SEO-PR mention tracking, and Site Audit for technical health. Pairs with the catalog's seo-keyword, seo-competitor, seo-content-gap-audit, content-strategy, programmatic-seo, and seo-onpage skills. Use alongside Ahrefs for cross-validation, especially on US-market keyword work where Semrush data depth often beats Ahrefs.

Official MCPSemrush official MCP, API key auth, paired with the catalog's SEO foundation skills.

What you can do via MCP

Example prompts the agent runs.

  • “Cluster these 200 seed keywords by topic and intent; return the 5 to 8 clusters that ladder up to a content hub.”

    Calls the Keyword Magic Tool plus Topic Research, runs intent classification per keyword, clusters by SERP overlap and topical similarity, returns the cluster structure with seed-piece recommendations. The pillar-content-architecture skill picks up these clusters as the input for hub-and-spoke planning.

  • “On the keyword 'experimentation analytics,' which competitors hold the answer-box, knowledge panel, and People Also Ask placements? Where do we show up in the broader SERP universe?”

    Calls Position Tracking and SERP Features, returns SERP-feature ownership per top-10 result, our current SERP-feature presence (or absence), and the gap. The seo-competitor skill's SERP overlap angle (#1) consumes this directly.

  • “Build a content brief for 'how to design an A/B test' targeting commercial-investigation intent; pull SERP-feature targets, related questions, and the entities every top-10 page covers.”

    Calls Topic Research for related questions and entity coverage, Position Tracking for SERP features the top 10 hold, and Keyword Magic Tool for variant queries. Returns a brief object the content-brief-authoring skill picks up: target keyword, intent, SERP-feature targets, entity list, related questions, recommended subsections.

  • “Track brand mentions of 'rampstack' across the web in the last 30 days; flag any that are not yet linked back to us, sorted by domain authority.”

    Calls Brand Monitoring, filters mentions to unlinked references in the time window, joins to source-domain authority, returns the prioritized outreach list. The seo-backlink-audit skill consumes this for unlinked-mention reclamation work.

  • “Compare our keyword universe to airbnb.com on the US market; surface keywords driving over 1,000 monthly searches that they rank for and we do not.”

    Pulls our ranking universe and the competitor's, intersects on keyword universe minus our rankings, filters to volume threshold, returns the gap list. seo-keyword-gap-audit picks up this list as Category 1 (Missing topics) input.

Semrush · MCP

Topic Research and Keyword Magic Tool query via the Semrush MCP. The agent specifies the seed keyword and target market; Semrush returns the keyword universe with SERP-feature presence and entity coverage.

Semrush MCP
docs
# Semrush MCP, called via the agent runtime
mcp.semrush.keyword_magic_tool({
  keyword: "experimentation analytics",
  database: "us",
  match_type: "phrase",
  filters: {
    search_volume_min: 100,
    keyword_difficulty_max: 60,
    intent: ["commercial-investigation", "informational"]
  },
  include_serp_features: true,
  limit: 50
})

# Returns: keywords[] with volume, difficulty, intent, cpc,
#          serp_features[], parent_topic, current_top_10_domains.
One command sample showing how the agent talks to Semrush. The MCP exposes the platform's primitives as tools; the agent translates the prompt into the right call.

MCP integration

Semrush MCP server.

Server
Semrush official MCP endpoint (verify URL at authoring time; Semrush ships first-party MCP tooling)
Auth
API key (per-account, rotatable in Semrush settings; subscription-tier scoping applies)
Hosting
Semrush hosted
Scoping note
MCP access is scoped to the API tier on the Semrush subscription. Pro and Guru tiers cover most catalog skill workflows; Business is needed for high-volume Position Tracking projects and unlimited Topic Research queries. The agent runtime respects API-call limits and surfaces remaining quota when it nears the cap.
  • Topic Research and Keyword Magic Tool for cluster expansion
  • Position Tracking with daily SERP-position and SERP-feature deltas
  • Brand Monitoring for SEO-PR mention tracking and unlinked-mention reclamation
  • Backlink Analytics with referring-domain authority and toxic-score flags
  • Site Audit for technical health (Core Web Vitals, crawl, schema)
  • Keyword Gap with multi-database support (US, UK, AU, DE, others)

Semrush API docs

Visual demonstration

What this looks like in practice.

Semrush· DashboardLast 30 days
Position Tracking · rampstack.co

  • Authority Score

    54

    +1.9%vs last 30d

  • Tracked Keywords

    320

    flatvs last 30d

  • Top 3 Positions

    42

    +5.0%vs last 30d

  • Top 10 Positions

    118

    +3.5%vs last 30d

  • SERP Features Won

    27

    -7.4%vs last 30d

  • Estimated Traffic

    31.4K

    +2.8%vs last 30d
The Position Tracking project the agent reads. The SERP-features-won regression (down 7.4 percent) is the prompt the seo-competitor skill picks up next: which features did we lose, and to whom?

CLI alternative

Semrush REST API for batch workflows.

The MCP is the primary surface for agent-driven SEO work. The REST API fills in for high-volume operations the agent runtime is not built for: scheduled exports of full Position Tracking history, bulk Keyword Magic Tool queries for programmatic-SEO seeding, and CI runs that page on Authority Score regressions.

Pairs with these skills

The SEO foundation skills plus content production.

Semrush pairs primarily with the SEO foundation skills (seo-keyword, seo-competitor, seo-onpage) and with content production (content-strategy, content-brief-authoring, programmatic-seo). Topic Research feeds brief generation; Position Tracking feeds ranking diagnostics; Brand Monitoring feeds SEO-PR work the audit-suite skills do not cover.

On Ahrefs, honestly.

Semrush and Ahrefs are siblings, not replacements. Both cover keyword research, backlinks, position tracking, and site audit. Which one wins depends on the workload. Semrush has stronger US keyword data, Topic Research, and SEO-PR features (Brand Monitoring, mention tracking) that Ahrefs does not match. Ahrefs has stronger global backlink data, a larger SERP archive, and tighter integration into the audit-suite skills as the assumed backend. Most teams running both run them in cross-validation mode: methodology decisions become higher-confidence when independent platforms agree, and disagreement is itself a useful signal that one platform is missing data the other captures.

Cost considerations

Semrush's API charges per call. Long agent runs can compound.

Credits per call

This MCP wraps a paid API. Each tool call consumes credits or units against your plan. In long agentic sessions, costs can compound: a Claude agent running "audit competitors X, Y, Z across keyword, backlink, and traffic dimensions" may burn hundreds of API calls before it returns a single deliverable.

Semrush API call costs vary by endpoint; some endpoints cost 10x others. Long agentic sessions can burn API budget quickly. Cache aggressively in agent workflows.

Recommended patterns

  • Cache responses in your agent's session memory rather than re-fetching the same data.
  • Scope queries narrowly before broad sweeps (a single competitor before all competitors).
  • Monitor your platform's usage dashboard during long workflows; set spend alerts where the platform supports them.
  • Batch where the API supports it; many endpoints accept arrays of inputs at the same credit cost as single inputs.
  • Endpoint cost varies sharply: a Domain Overview call is cheap, a full Backlink Analytics export is expensive. Read the API Units tariff before authoring agent workflows.
  • Use Position Tracking projects rather than ad-hoc rank lookups; project-based queries amortize their setup cost across daily reads.

Continue reading.