Developers

Build on RankSage — MCP first.

RankSage's programmatic surface is a read-only MCP server over the whole product — AI visibility, content intelligence, competitors, behaviour, traffic and the action queue — plus the OAuth 2.0 flow that authorizes it. Everything on this page has a machine-readable twin, so an agent can integrate without a human reading any of it.

The RankSage MCP server

A Model Context Protocol server speaking Streamable HTTP, exposing 17 read-only tools. It works out of the box in Claude, Claude Code and Cursor: add the server URL, and the client discovers the OAuth flow automatically from the published metadata. All data access is tenant-bound server-side — no tool accepts a user or website identifier, so a token can only ever read its own account.

Server URL:  https://web-production-a039f.up.railway.app/api/v1/mcp
Transport:   streamable-http
Auth:        OAuth 2.0 (PKCE S256, dynamic registration) or personal access token

get_digest

The cross-surface RankSage digest: visibility, traffic, content and action highlights for a period.

get_visibility_summary

AI visibility summary — citation rates and share of voice across the tracked answer engines.

get_visibility_detail

Per-engine, per-prompt AI visibility detail with run history.

get_ai_performance

AI performance rollup — AI-referred traffic joined to visibility.

get_search_analytics

Classic search analytics: Search Console queries, top pages and PageSpeed.

get_traffic_summary

GA4-derived traffic summary with engagement metrics.

get_content_report

Content quality scores, E-E-A-T and citability per crawled page.

get_content_production

Content production status — briefs, drafts and optimizer runs.

get_gap_counts

Content gap counts versus tracked competitors.

list_gaps

Content gaps with the competitor evidence behind each.

get_competitor_report

Competitor crawl scores, rank matrix and citation benchmarks.

get_community_report

Forum/Reddit SERP presence and community mentions.

get_behavior_report

First-party behaviour: frustration signals, heatmap and session rollups.

get_site_readiness

Technical + agent readiness checks for the connected site.

list_action_queue

The ranked action queue merging recommendations across surfaces.

list_conversationsscope: conversations

List recent RankSage assistant conversations.

get_conversationscope: conversations

Read one stored assistant conversation.

Authentication and scopes

OAuth 2.0 authorization-code with mandatory PKCE (S256) and RFC 7591 dynamic client registration — no manual credential issuance. Headless agents can instead use a personal access token created in the app (Settings → MCP), scoped at creation and stored hashed. The full walkthrough, written for agents, is at /auth.md.

mcp

Read-only access to every core RankSage MCP tool: visibility, content, competitors, behaviour, traffic and the action queue. No write operations exist on this surface.

conversations

Read stored RankSage assistant conversations. Grantable only to personal access tokens created in the RankSage app with explicit consent — OAuth grants cannot carry this scope.

RankSage agent & developer files

Every discovery file this domain publishes, at its standard path.

/openapi.json
OpenAPI 3.1 spec. The API in machine-readable form, including OAuth security schemes and scopes.
/.well-known/mcp/server-card.json
MCP server card. Name, transport, auth and the full tool roster — preview before connecting.
/.well-known/ai-catalog.json
AI resource catalog (ARD). One catalog of everything this domain offers agents.
/.well-known/api-catalog
API catalog (RFC 9727). IETF-standard linkset pointing at the API and its descriptions.
/.well-known/oauth-authorization-server
OAuth authorization server metadata (RFC 8414). Endpoints, grant types, PKCE methods, scopes.
/.well-known/oauth-protected-resource
OAuth protected resource metadata (RFC 9728). What the MCP endpoint protects and which scopes it understands.
/auth.md
auth.md. A step-by-step credential walkthrough written for agents.
/llms.txt
llms.txt. Structured product summary for language models (full text: /llms-full.txt).
/agents.md
agents.md. Operating instructions for autonomous agents visiting this site.
/pricing.md
pricing.md. Machine-readable pricing tiers and limits.

Versioning and deprecation

The API is versioned in the URL path (/api/v1). Breaking changes ship as a new path version, never in place; when a version is scheduled for retirement, the deprecation is announced on this page and in the OpenAPI spec with at least 90 days of overlap in which both versions work. Additive changes (new tools, new optional fields) arrive within the current version.

The API itself is read-only and rate-limited; there are no write endpoints on the agent surface. Questions or something broken? hello@ranksage.com.

Your AI-visibility data, inside your agent.

Connect the MCP server and ask your assistant what changed this week — with numbers traceable to stored runs.