Skip to content

Changelog

Every Agent API and MCP server release, newest first.

The current API version is returned in the X-API-Version header of every response. The base path stays /api/agent/v1; the minor version moves as endpoints and fields are added. Release notes for the whole Rankability app are in the help center changelog.

November 6, 2026

AI assistants - Connect and manage your assistants from the sidebar

Section titled “AI assistants - Connect and manage your assistants from the sidebar”

You can now connect Rankability to the AI assistant you already use from the sidebar, then see every connection in one place.

Connect where you work — Choose AI assistants in the sidebar, pick your assistant, and follow its setup steps. Rankability then works inside that assistant with your organization’s permissions.

See what is connected — The list shows each connected assistant and when it was last used. Assistants with more than one connection are grouped so you can review them together.

Remove access you no longer need — Disconnect an assistant from its menu after a confirmation. Settings → Connected apps remains available for the same task.

Follow Connecting Rankability to AI assistants with MCP to connect your first assistant.

October 21, 2026

Help Center - Give your AI assistant current Rankability instructions

Section titled “Help Center - Give your AI assistant current Rankability instructions”

You can now connect an MCP-compatible AI assistant directly to the public Rankability Help Center, so it can find product instructions while you work. No Rankability account or API key is needed for this documentation connection.

Find the right guide — Ask a practical question, such as how to choose platforms for keyword tracking. The connector searches the published guides and returns links to the relevant sections.

Read the instructions in context — Your assistant can retrieve a complete article or browse the available sections before explaining the workflow. Follow the source links to check its answer against the documentation.

Keep the connection scoped — This connector reads public documentation only. Access to your clients, reports, scans, and other account actions requires the separate authenticated Rankability connection.

Follow Use Help with your AI assistant to connect.

August 28, 2026

MCP v1.15 - Do more Rankability work from your AI assistant

Section titled “MCP v1.15 - Do more Rankability work from your AI assistant”

Rankability’s Agent API and hosted MCP connection now cover substantially more of the platform, so an approved AI assistant can inspect saved work, start supported jobs, and follow results without losing the client context.

Work across the platform — The MCP server now registers 90 tools spanning clients, Knowledge, Researcher, Copywriter, Optimize, Tracker, Search Console, auditing, crawling, Search Intelligence, Prospector, backlink evidence, Routines, publishing, and Serena consultation.

Read saved evidence before starting work — Your assistant can inspect existing projects, results, usage history, audit artifacts, outreach lists, and publishing receipts before proposing another run. Missing measurements remain unavailable rather than becoming false zeros.

Keep consequential actions controlled — OAuth and scoped API keys limit access. Work that uses pooled capacity requires an impact estimate and your confirmation, while retry-safe operations use a stable request key to avoid duplicate jobs or changes.

Follow Connecting Rankability to AI assistants with MCP to reconnect, verify the active organization, and load the current tool set.

August 14, 2026

Rankability - Connect Claude or Codex and work through your AI assistant

Section titled “Rankability - Connect Claude or Codex and work through your AI assistant”

Rankability now provides a guided way to connect Claude or Codex and use the selected client’s Rankability context and tools from your assistant.

Start from organization settings — Open Settings → API keys, choose Connect Claude or Codex, and follow the assistant-specific OAuth or scoped-key instructions.

Verify the connection before starting work — The setup checks that the Rankability endpoint is ready and gives you a read-only verification prompt. Confirm the active organization, shared-credit balance, and available tools before approving a consequential action.

See whether the connection is actually ready — Connection status distinguishes a configured assistant from one that still needs authentication or attention, reducing guesswork when no tools appear.

Keep usage and job states clear — The assistant can read the active credit state, estimate supported work before it starts, and distinguish saved drafts, queued jobs, stale integrations, and unavailable measurements instead of presenting them as completed work or zeros.

Follow Connecting Rankability to AI assistants with MCP to connect and verify your assistant.

July 7, 2026

API v1.8 - Run Copywriter Autopilot from the API and AI assistants

Section titled “API v1.8 - Run Copywriter Autopilot from the API and AI assistants”

The new hands-off Copywriter Autopilot is also available programmatically — start a full topic-to-finished-draft run from your own systems or from an AI assistant connected over MCP.

Start a run with one call — POST /api/agent/v1/copywriter/autopilot takes a topic (plus optional client, intent, and extra context) and kicks off the entire pipeline: research, brief, draft, optimization passes, fact-checking with citations, and imagery. You get a job ID back immediately and the run continues in the background.

Poll progress stage by stage — GET /api/agent/v1/copywriter/autopilot/:id reports exactly where the run is — researching, briefing, drafting, optimizing, fact-checking, adding imagery, finalizing — plus how many optimization passes have run and whether the flat 3,000-credit charge has been applied or released.

Safe to retry — Send an Idempotency-Key header and retrying the same request returns the original job instead of starting (and charging for) a duplicate run.

Fetch the finished draft — When the run completes, pull the final content, brief, and images through the same artifacts endpoint used by other content jobs.

Built in for AI assistants — Connect over MCP and your assistant can start and monitor runs directly with the new create_content_autopilot and get_autopilot_status tools.

Generate or rotate a key in Settings → API Keys, and see the full endpoint reference in the Help Center.

July 3, 2026

Copywriter API - Cheaper default and clear cost on every job

Section titled “Copywriter API - Cheaper default and clear cost on every job”

The Copywriter API now defaults to a cheaper, recommended set of research platforms and tells you what each job will cost — so automated content pipelines stop burning credits they didn’t mean to.

A cheaper default when you don’t choose — If you create a Copywriter job without specifying research_platforms, the API now runs the recommended subset — Google Organic, ChatGPT, and Claude — the same default the in-app picker uses, instead of running every platform. For a full auto run that’s roughly 1,036 credits instead of ~2,100, with no change to how you call the API.

You’re always in control — Want more sources? Pass a research_platforms array and the API runs exactly what you list (Google Organic is always included). Passing the full list runs every platform, just like before.

See the cost before the run finishes — Every job-creation response now echoes back the research_platforms actually used and a credits_estimated figure, so your automation can log and budget the exact cost of each run.

Heads up if you relied on the old behavior — If your integration expected every platform to run when research_platforms was omitted, add the full array explicitly to keep that behavior. Otherwise, you’ll simply spend fewer credits per run.

See the updated field reference and cost guidance at docs/agent-api.md.

June 14, 2026

API v1.7 - Run full-site audits via the API and MCP

Section titled “API v1.7 - Run full-site audits via the API and MCP”

The Site Auditor is now available through the API and to AI assistants over MCP, so you can crawl an entire site and pull back per-page technical health programmatically — not just from the app.

Start a full-site crawl — POST /api/agent/v1/site-auditor/projects creates a project and kicks off a crawl of the whole site, then POST /api/agent/v1/site-auditor/projects/:id/crawl re-runs it whenever you want fresh data. Crawls run in the background, so you create the project and poll for results.

Stop a crawl mid-run — Changed your mind or pointed at the wrong site? POST /api/agent/v1/site-auditor/projects/:id/cancel aborts an in-progress crawl, and a cancelled run is never charged — you only pay for crawls that finish.

Read the findings — List your projects, or fetch one to get its pages plus site-wide issues — broken internal and external links, duplicate content, indexability, and canonical clusters — the same findings you see in the in-app Site Auditor.

Know the cost before you crawl — A credit estimate endpoint previews the cost for a given page count, and crawls are charged on success at 1 credit per 5 pages crawled, so you only pay for what actually completes.

Clean up when you’re done — Finished with a project? DELETE /api/agent/v1/site-auditor/projects/:id permanently removes it along with all of its crawled pages and issues, so your project list stays tidy. An in-progress crawl is protected — cancel it first, then delete.

Built in for AI assistants — Connect over MCP and your assistant can run, cancel, read, and clean up crawls directly with the new site_audit_run, site_audit_cancel, site_audit_get, site_audit_list, and site_audit_delete tools.

Generate or rotate a key in Settings → API Keys, and see the Site Auditor endpoints documented at docs/agent-api.md.

June 5, 2026

API v1.6 - Choose research platforms for Copywriter jobs

Section titled “API v1.6 - Choose research platforms for Copywriter jobs”

The Copywriter API now lets you choose exactly which research platforms power a draft — so you can control both research depth and credit cost on every request.

A new research_platforms parameter — POST /api/agent/v1/copywriter/jobs now accepts an array of the SERP and AI sources you want the run to use. Google Organic stays on as the baseline, and the Copywriter only researches the platforms you list — skip the ones you don’t need for leaner, faster jobs.

Pay only for what you use — Credit cost now scales with the platforms you select, and the up-front credit check reflects your choice before the job starts, so trimming the list trims the cost of each run.

Matches the in-app picker — These are the same research platforms you can toggle when creating a Copywriter project in the app, so jobs created through the API and the UI behave consistently.

Generate or rotate a key in Settings → API Keys, and see research_platforms documented at docs/agent-api.md.

May 8, 2026

API v1.5 - OAuth 2.0 on remote MCP, Connected apps in Settings, and n8n on the Integrations page

Section titled “API v1.5 - OAuth 2.0 on remote MCP, Connected apps in Settings, and n8n on the Integrations page”

Connecting Rankability to AI assistants and automation tools just got a lot more grown-up. The remote MCP endpoint now speaks standards-compliant OAuth 2.0, you can review and revoke every authorized app from your settings, and the n8n community node is listed front-and-center on the in-app Integrations page.

OAuth 2.0 on the remote MCP endpoint — Connecting Claude (or any other MCP client) to Rankability now goes through a standard OAuth flow with an explicit consent screen. You’ll see exactly which permissions the client is asking for — for example Create and run content generation jobs or Query search intelligence (Google + AI platforms) — and you choose Allow Access or Deny before any token is issued. Discovery endpoints (/.well-known/oauth-authorization-server, /.well-known/oauth-protected-resource) are live so compliant clients can connect with zero manual setup.

Connected apps tab in Settings — A new Connected apps tab in Settings lists every app you’ve authorized through OAuth (Claude, custom MCP clients, third-party assistants), the date you authorized it, when it was last used, and the exact scopes it has. Each one has a Revoke button with a confirmation dialog — kill access in one click without rotating any keys.

n8n node on the in-app Integrations page — The Settings → Organization → Integrations tab now features the official n8n community node card with one-line install instructions (search n8n-nodes-rankability in n8n, paste your API key) and three copy-pasteable starter workflows: Copywriter from Google Sheets, Weekly Tracker Digest to Slack, and Bulk Page Auditor from URL List. Connect Rankability to 400+ apps without writing code.

Open Settings → Connected apps to manage authorized assistants, or jump to Integrations to grab the n8n node.

April 27, 2026

API v1.4 - Researcher jobs refund credits on failure (and tell you about it)

Section titled “API v1.4 - Researcher jobs refund credits on failure (and tell you about it)”

When a Researcher job submitted through the Agent API fails, the credits charged at the start of the run are now refunded to your account automatically — and the API tells your integration exactly how much came back.

Automatic refund when a Researcher job fails — POST /api/agent/v1/researcher/jobs charges credits up-front when it queues a research run. If the worker (or the inline fallback) ultimately fails, the full charge is reversed back into your balance. Mixed charges that straddle included-plan credits and purchased credits are returned to the right buckets so your balance stays accurate.

credits_refunded on the job-detail endpoint — GET /api/agent/v1/researcher/jobs/:job_id now returns a credits_refunded field. For failed jobs it shows the exact number of credits that were returned (or 0 if the failure happened before the charge). Pending, processing, and completed jobs always return 0. No need to inspect transactions to confirm a refund landed.

Idempotent and safe to poll — The refund is keyed off the job id, so even if your integration retries or reconciles repeatedly, credits only come back once. Cache hits and pre-charge failures correctly report 0 refunded.

Generate or rotate a key in Settings → API Keys, and see the new field documented at docs/agent-api.md.

April 25, 2026

API v1.3 - Researcher endpoints for keyword jobs and saved projects

Section titled “API v1.3 - Researcher endpoints for keyword jobs and saved projects”

The Agent API now covers Researcher. Start keyword research jobs from your own tooling, poll for results, list past jobs, and read any saved Researcher project with its full keyword set — all with the same auth, rate limits, and audit logging as the rest of the API.

Start keyword research jobs — POST /api/agent/v1/researcher/jobs kicks off a research run in any of the supported modes (discover, trending, reddit, youtube, ecommerce). Jobs run asynchronously and return a job id you can poll. Cache hits return synchronously and cost zero credits.

Poll status and list past jobs — GET /api/agent/v1/researcher/jobs/:job_id returns live status and the result payload when the job finishes. GET /api/agent/v1/researcher/jobs lists every research job your account has run, with filters for status and mode.

Read saved Researcher projects — GET /api/agent/v1/researcher/projects lists every saved keyword project on your account. GET /api/agent/v1/researcher/projects/:id returns a single project with its full keyword set, volumes, intent, and clusters — ready to drop into a brief, an outreach list, or your own scoring pipeline.

Two new scopes — Issue keys with researcher:run to start jobs and researcher:read to read jobs and projects. Both scopes show up in the Settings → API Keys picker.

Up-front credit costs — Discover mode costs 500 credits, trending/reddit/youtube/ecommerce cost 750. Charges happen at job start, mirroring the Optimize pattern.

Generate a key in Settings → API Keys and read the full reference at docs/agent-api.md.

April 8, 2026

MCP now includes an optimize_page tool, letting your AI assistant score any public URL against its SERP competitors for a keyword — returning a full Rankability score, entity coverage analysis, and competitor benchmarks.

Score pages from your AI assistant — Ask your AI assistant to analyze a page and it calls the Optimize tool directly. Send a URL and keyword, and get back a 0–100 Rankability score based on entity coverage and keyword placement, a complete list of extracted entities with importance rankings, and a summary of the competitor pages that were analyzed.

New scope required — The tool requires the optimize:run scope on your API key. If you created your API key before this update, go to Settings → API Keys and add the optimize:run scope to enable the tool.

Same pipeline as the app — The MCP tool runs the exact same analysis pipeline as the in-app Optimize mode: SERP and Brave competitor scraping, GPT entity extraction, source URL scoring. Results are identical whether you use the app, the API, or MCP.

1,200 credits per call — Each optimize call costs 1,200 credits, consistent with Copywriter Quick mode pricing. Retry-safe with idempotency key support.

Update your MCP configuration and add optimize:run to your API key scopes to get started. See the MCP guide for setup instructions.

April 8, 2026

The Rankability API now includes a public Optimize endpoint that scores any URL against its SERP competitors for a given keyword — returning a full Rankability score, entity coverage analysis, and competitor benchmarks in a single API call.

One call, full analysis — Send a POST request with a URL and keyword to POST /api/v1/optimize and get back everything: a 0–100 Rankability score based on entity coverage (how many important entities from competitors your content includes) and keyword placement (whether your primary keyword appears in key spots like the title, H1, first paragraph, subheadings, and meta description), a complete list of extracted entities with importance rankings, and a summary of the competitor pages that were analyzed.

No project required — Unlike the Copywriter API, the Optimize endpoint is completely stateless. It doesn’t create a database project, doesn’t require a client ID, and doesn’t need polling. You send a request and get the full result in the response. This makes it ideal for batch scoring, content audits, and integration into external pipelines.

Entity extraction from competitors — The endpoint scrapes up to 20 SERP competitors and extracts the entities (brands, concepts, features, terms) that top-ranking pages cover. Each entity includes an importance level based on how many competitors mention it: required (5+), recommended (2–4), or optional (1). Use this to identify exactly which topics and entities your content is missing.

Built-in security — The endpoint blocks requests to private networks, localhost, cloud metadata endpoints, and internal hosts. URL validation ensures only public HTTP/HTTPS pages are analyzed.

Credit-efficient — Each call costs 1,200 credits (same as Copywriter Quick mode). Repeated requests for the same URL and keyword combination within a billing cycle are not double-charged thanks to automatic idempotency.

Authentication — The Optimize endpoint uses admin key authentication via the X-Admin-Key header. This is separate from the rk_live_ Bearer token system used by other API endpoints.

Full endpoint documentation is available in the Help Center under “Optimize API endpoint.”

April 1, 2026

The Rankability API now includes a Knowledge Base endpoint that gives you programmatic access to every help article and FAQ in the platform — so your AI agents, internal tools, and support systems can stay in sync with the latest Rankability documentation.

Full article access — The new GET /api/agent/v1/kb/articles endpoint returns every published Knowledge Base article with its title, content, category, and metadata. Use it to feed Rankability documentation into your AI assistants, internal wikis, or client-facing help portals.

JSON and Markdown formats — Request articles as structured JSON (default) for programmatic consumption, or set Accept: text/markdown to get a single Markdown document organized by category — ideal for feeding directly into LLMs or documentation generators.

Efficient caching — The endpoint supports conditional requests with ETag and Last-Modified headers. Your integration can skip re-downloading content that hasn’t changed, reducing bandwidth and API calls. Send If-None-Match or If-Modified-Since headers and receive a 304 Not Modified when nothing has been updated.

FAQ included — The response includes the full Rankability FAQ alongside the Knowledge Base articles, giving your tools a complete picture of platform capabilities, pricing, and common questions.

New scope — Add the kb:read scope to your API key to access this endpoint. You can add it to an existing key or create a dedicated one from Settings > API keys.

Use cases — Train an internal AI support agent on Rankability’s full documentation. Sync help content into your agency’s knowledge management system. Build a custom help widget that pulls live content from Rankability. Power an onboarding chatbot that answers new team members’ questions about the platform.

Full endpoint documentation is available in the Help Center under “Knowledge Base API endpoint.”

March 19, 2026

MCP v1.0 - Connect Rankability to your AI assistant

Section titled “MCP v1.0 - Connect Rankability to your AI assistant”

Rankability now supports the Model Context Protocol (MCP), letting you connect your SEO data directly to AI assistants like Claude Desktop, Cursor, and Windsurf. Ask your AI assistant about your clients, rankings, and content — and it can read and act on your Rankability data in real time.

Works with your favorite AI tools — Connect Rankability to Claude Desktop, Cursor, Windsurf, or any MCP-compatible AI assistant. Your assistant gets access to your clients, content projects, rank tracking data, and credit balance through a secure, authenticated connection.

Read your SEO data conversationally — Ask your AI assistant questions like “How is Acme Corp performing?” or “Which keywords dropped this week?” and it pulls live data from Rankability. List clients, check SPI scores, review content project status, and browse ranking results — all through natural conversation.

Create content and trigger scans — Your AI assistant can create new content jobs, approve research briefs, and trigger rank tracking scans on your behalf. Use auto mode for fully hands-off content generation, or stepped mode to review the brief before generating the full draft.

Two connection methods — Use the hosted HTTP endpoint for Cursor and Windsurf (just add the URL and your API key to your MCP config), or connect Claude Desktop through the OAuth connector (Settings → Connectors → Add custom connector — no API key or local install needed). Setup takes under a minute either way.

Secured by your API key — MCP connections use the same API keys you already manage in Settings. All the same scopes, rate limits, and credit controls apply. Your data stays within your organization’s access boundaries.

Get started by visiting the MCP integration guide in the help center.

March 16, 2026

API v1.1 - Full platform access and abuse protection

Section titled “API v1.1 - Full platform access and abuse protection”

The Rankability API now covers the entire platform — not just content creation. You can manage clients, tracking projects, and content programmatically, with built-in protection against runaway integrations.

Client management — Create, update, and delete clients via the API. Set brand voice, website URL, industry, and custom instructions when creating a client, or update them later. Use clients:read and clients:write scopes to control access.

Tracker tracking — Create tracking projects with keywords, platforms, and locations. Update project settings, trigger on-demand scans, and pull ranking results and trends — all through the API. New scopes: reporter:write for creating and modifying projects, reporter:run for triggering scans, and reporter:read for pulling data.

Copywriter list, export, and delete — List all your content projects with filtering, export finished articles as HTML, Markdown, or Word, and delete projects you no longer need. These join the existing job creation and status endpoints.

Abuse protection — The API now uses a sliding-window rate limiter that distributes requests evenly instead of allowing bursts at window boundaries. Keys that generate excessive errors or rate limit violations are automatically suspended with a clear error message explaining the suspension and how to resolve it. All rate-limited requests are now logged for full visibility. Your admin can reactivate suspended keys from the API settings page.

API version header — Every API response now includes an X-API-Version header so your integrations can detect which version they’re running against. The current version is 1.1.0.

March 4, 2026

Tracker API v1.0 - Tracking data for your dashboards and tools

Section titled “Tracker API v1.0 - Tracking data for your dashboards and tools”

Your Tracker data is no longer locked inside the app. The new Tracker API gives you programmatic access to every SPI score, platform result, and trend in your account — so you can pipe it into the tools your clients and team already use.

Build client-facing dashboards — Pull SPI scores, trend history, and platform breakdowns into Looker Studio, Google Sheets, or any BI tool. Create live dashboards that update automatically, branded with your agency’s look, and share them with clients who want to see progress without logging into another platform. No more screenshotting charts or copy-pasting numbers into slide decks.

Monitor all your clients from one place — The summary endpoint returns your organization-wide average SPI, top movers (biggest gains and drops), and platform coverage across every project. Use it to build an internal command center that flags which clients need attention and which are trending up — without clicking through dozens of individual projects.

Trigger scans on your schedule — Need fresh data before a client call or a Monday morning report? The scan endpoint lets you kick off Tracker scans programmatically. Pair it with a cron job or automation platform to keep your dashboards current on exactly the schedule you want.

Same authentication, same keys — The Tracker API uses the same API keys and Bearer token format as the Copywriter API. Add the reporter:read scope to fetch data, and reporter:run if you also want to trigger scans. You can add both reporter scopes to an existing key or create a dedicated one — your call.

Six endpoints, everything you need — List projects with filtering by client and status. Get a single project’s detail with its full SPI breakdown (traditional ranking, AI mentions, AI citations). Fetch per-platform results from any historical run. Pull up to 90 days of SPI trend data for time-series charts. Trigger new scans. And get an aggregate summary across your entire portfolio.

Looker Studio guide included — We’ve published a step-by-step guide in the Help Center showing how to build a Google Apps Script community connector, map API responses to Looker Studio data sources, and design a dashboard layout with SPI scorecards, trend charts, and project tables. It covers caching, rate limit management, and multi-client setups for agencies managing many accounts.

To get started, go to Settings > API keys to create or update a key with reporter scopes. The quick start guide on that page now includes Tracker API examples alongside the Copywriter API. Full endpoint documentation is available in the Help Center.

February 25, 2026

You can now create SEO content programmatically through our new REST API. Connect tools like OpenClaw, custom scripts, or automation platforms to Rankability and let them run the full Copywriter pipeline — research competitors, generate briefs, and produce optimized drafts — all without opening the app. The API supports everything you need: custom AI instructions to guide the writing, location and language targeting for geo-specific content, tone presets, auto mode for hands-off generation, and stepped mode when you want to review the brief before drafting. Every job returns the finished article along with SEO metadata, FAQ schema, entity coverage, and a quality score. The Agent API is available on every plan — Core, Team, and Agency. Higher plans unlock faster rate limits and priority queue processing so your jobs complete sooner. To get started, go to Settings > API keys to generate your first key and copy the ready-made instructions for your AI agent.