MCP server
Connect Claude, ChatGPT, Cursor, Codex or any MCP client to Rankability's hosted MCP server, and see every tool it exposes.
Rankability hosts a remote Model Context Protocol server. There is nothing to install: point an MCP client at the endpoint, sign in, and the assistant gets the same tools the Agent API offers, scoped to one Rankability organization.
https://app.rankability.com/mcp- Transport: Streamable HTTP.
- Sign-in: OAuth 2.0 with PKCE (preferred), with dynamic client registration. Metadata is at
https://app.rankability.com/.well-known/oauth-authorization-server. - API key: clients that cannot do OAuth can send
Authorization: Bearer rk_live_...with a scoped key from Settings > API keys. - Limits: every tool call goes through the Agent API with the caller’s credential, so it uses the same scopes, the same per-key rate limit and the same usage as a direct API call.
Connect a client
Section titled “Connect a client”Claude
Section titled “Claude”Rankability is listed in Claude’s connector directory. In Claude, open Customize > Connectors > Discover, search for Rankability, select Connect to Claude, then sign in and approve access. On Team and Enterprise plans an owner connects it for the team first.
Add the server under Settings > MCP servers > Add server as Streamable HTTP, or in ~/.codex/config.toml:
[mcp_servers.rankability]url = "https://app.rankability.com/mcp"auth = "oauth"default_tools_approval_mode = "writes"Cursor, VS Code, Windsurf and other clients
Section titled “Cursor, VS Code, Windsurf and other clients”Add a remote server with the endpoint above. Choose OAuth when the client offers it. Otherwise use a key:
{ "mcpServers": { "rankability": { "url": "https://app.rankability.com/mcp", "headers": { "Authorization": "Bearer rk_live_YOUR_API_KEY" } } }}Give that key only the scopes the assistant needs, and keep the file out of version control.
Step-by-step setup for app users, including ChatGPT and v0, is in the help center: Connecting Rankability to AI assistants with MCP.
Try it
Section titled “Try it”Start with a read-only prompt. It confirms the connection without starting work:
Use Rankability to show my organization and usage limits without starting any work.Then ask for something useful, for example:
Use Rankability to show where Acme Plumbing is missing from AI answers and top search results, and which pages I should target first.How the tools behave
Section titled “How the tools behave”- Reads are free to run and safe to retry. Tools marked read-only never start scans, jobs or charges.
- Work that uses your plan asks first. Tools that run scans, research, audits or content first need a usage estimate (
estimate_usage_impact, orsite_audit_estimatefor crawls), and the call itself must carryconfirm_usage: true. Well-behaved clients show the estimate and ask the user before confirming. - Retries do not duplicate work. Creation tools accept an
idempotency_key; reuse it when you retry. - Errors are readable. A failed call returns the API’s error code and a message an assistant can relay to the user as is.
The server exposes 127 tools to a regular organization, grouped below. The list is generated from the server’s own tool definitions. Some organizations see a few extra pilot tools.
Serena
Section titled “Serena”| Tool | What it does | Access | Billing |
|---|---|---|---|
consult_serena |
Use this when the user asks a strategic question about one client, such as what to prioritize next or why visibility changed, and wants an answer grounded in that client’s saved Rankability data. Ask Serena a client-scoped strategic question. Serena reads the client’s selected knowledge sources, saved workspace evidence, durable client memory, Rankability product documentation, and Rankability methodology, then returns one grounded answer. This tool is read-only: it never runs scans, changes projects, publishes, or performs live web research. Do not use for live web research or to run scans. | Read-only | Plan only |
get_knowledge_document_review |
Read a saved review status, inspected coverage and completed answer. Edited or unavailable sources make the result stale; incomplete reviews never claim full input coverage. No new model work or usage is started by this reader. | Read-only | Plan only |
review_knowledge_document |
Start an explicitly requested, usage-bearing, background section-by-section review of original Knowledge text. No source editing, re-importing or publishing. Up to 20 sources / 500,000 characters total. Omit source_ids to review selected sources. Returns a saved review ID; queued/processing is NOT complete. Poll get_knowledge_document_review for the answer and verified coverage. Unchanged identical requests reuse the saved review. | Creates or updates | Plan only |
Search Console and analytics
Section titled “Search Console and analytics”| Tool | What it does | Access | Billing |
|---|---|---|---|
get_gsc_search_performance |
Use this when the user asks about Google Search Console clicks, impressions, CTR, queries or page performance. Read organization-scoped Google Search Console performance at query, page, or query-page grain. Defaults to the latest complete 28-day window, returns at most 100 rows per page, and supports a maximum 120-day range. Position is impression-weighted; empty data is not treated as zero demand. Do not use for AI answer visibility (use get_visibility_summary). | Read-only | Free |
Clients and Knowledge
Section titled “Clients and Knowledge”| Tool | What it does | Access | Billing |
|---|---|---|---|
create_client |
Create a client in the organization. Use create_client_connect_link afterward when the user wants to connect GSC or GA4. | Creates or updates | Free |
create_client_connect_link |
Create a secure, expiring browser link for a user to authorize Google and select GSC and GA4 properties for a client. Google consent still requires a human. Creating a new link revokes any older active setup link for this client. | Changes or deletes data | Plan only |
create_client_source |
Add one pasted-text source to a client’s Knowledge. When upsert_by_name is true, replace the existing exact-name source’s text with the supplied text. Use reference_only for internal research that must stay readable without entering automatic Copywriter prompts. Large sources remain one logical source and are chunked internally for indexing. The response reports current indexing state plus a durable contradiction-check receipt; poll get_client_source_contradictions when it is scheduled. This does not crawl a URL. | Changes or deletes data | Per call |
delete_client |
Permanently delete one client workspace and associated Tracker history. Requires the exact current client name as a second confirmation. Prefer preserving the workspace unless deletion is explicitly requested. | Changes or deletes data | Free |
delete_client_source |
Permanently delete one client Knowledge source. Requires the exact current source name as a second confirmation. | Changes or deletes data | Free |
get_client |
Get the lightweight client identity/configuration primitive: domain, integration health/freshness, structured business scope, settings, effective content brand/profile, and exclusion controls. Use get_client_overview when you also need Tracker, Copywriter, and Site Auditor state. | Read-only | Free |
get_client_overview |
Get a one-call, read-only client workspace overview combining identity and integrations, Tracker portfolio and SPI, recent Copywriter projects, and the latest Site Auditor project. Use this after resolve_client for broad questions such as ‘How are we doing?’. | Read-only | Free |
get_client_source |
Read one Knowledge source. Full view includes the complete approved source body and should be requested only when needed. | Read-only | Free |
get_client_source_contradictions |
Read durable contradiction-check receipts and completed findings for one Knowledge source. Use this after create_client_source or a content-changing update returns a scheduled check. | Read-only | Free |
get_client_source_section |
Get the complete section map and optionally one exact original section of a saved Knowledge source. Zero-based section index. Each section has a SHA-256, UTF-16 character range, UTF-8 byte range and line range. This read is not a full-document review; source completeness and review completeness are different. | Read-only | Free |
get_client_visibility_gaps |
Find where one client is missing, in one call, from existing Tracker results (no scan, no page fetch, no usage). Returns every ranking or AI-cited page across the client’s reports that does not mention the client, deduplicated by page, with page type (directory_listing, best_of_list, review_platform, info_page, community_thread, competitor_site, lead_gen_doorway, other), geo match to the client’s market, the competitors on the page, the platforms and number of prompts it shows up for, and a priority. Check ready before treating the list as final; while false, source_check_progress reports checked and pending work plus when to repeat the same call. Lead-gen doorways, wrong-city and off-topic pages are hidden unless include_excluded=true. Pages the app could not read (login walls, blocked pages) are marked unverified. Also returns AI answers that omit the client and queries where the client is outside the organic top 10 or the Local Pack top 3. Brand matching uses the brand name, aliases, and GBP location name, and saved page text corrects mention checks that missed an alias. Compact and paginated by default; view=full adds per-page evidence. | Read-only | Free |
import_client_source_file |
Import one UTF-8 Markdown or CSV file into a client’s Knowledge. Supply the file name and complete text content; the MCP client reads the local file and sends its text without exposing a local path to Rankability. When upsert_by_name is true, an exact file-name match has its existing text replaced by the supplied text. The response includes indexing state and a durable contradiction-check receipt. | Changes or deletes data | Per call |
list_client_sources |
List bounded client Knowledge sources. Compact view returns source state and character counts without source bodies. | Read-only | Free |
list_clients |
Use this when the user asks which clients or websites they manage in Rankability, or before any client-scoped request when no client is chosen yet. List all clients in the organization. Returns client names, domains, and connected integrations. | Read-only | Free |
resolve_client |
Resolve a human-readable client name or domain to organization-scoped client candidates. Use this before client-scoped tools when the user supplied a name or website instead of a UUID. If multiple matches are returned, ask the user to choose before taking consequential action. | Read-only | Free |
update_client_profile |
Patch client identity, approved workspace Tracker aliases, or structured brand/writing guidance. Knowledge fields are merged into the existing profile, so omitted sections and source documents are preserved. Per-location GBP names remain project settings managed with update_tracker_project. | Changes or deletes data | Free |
update_client_source |
Patch one Knowledge source’s name, approved body, role, generation usage, selection state, or priority. Omitted fields are preserved. | Changes or deletes data | Per call |
AI and search visibility tracking
Section titled “AI and search visibility tracking”| Tool | What it does | Access | Billing |
|---|---|---|---|
compare_visibility |
Compare one client’s visibility at two dates from existing Tracker results (no scan, no usage). Returns like-for-like AI mention and citation rate changes, cells that gained or lost a mention or citation, organic and local pack position changes, and new or lost competitors. Cells unmeasured on either side are excluded, not counted as losses. Compact by default; view=full returns every change. | Read-only | Free |
create_tracker_project |
Create one independent Tracker report for exactly one keyword. This does not run an immediate scan. Benchmark mode is the safe default; enable recurring tracking only after the user approves its schedule. | Creates or updates | Per call |
delete_tracker_project |
Permanently delete one Tracker report and its history. The caller must supply the exact current keyword as a second confirmation; use this only when preservation or archiving is not appropriate. | Changes or deletes data | Free |
get_reporter_project |
Legacy compatibility alias for get_tracker_project. Reporter is retained only in the tool ID and API namespace. | Read-only | Free |
get_reporter_summary |
Legacy compatibility alias for get_tracker_summary. Reporter is retained only in the tool ID and API namespace. | Read-only | Free |
get_tracker_brand_summary |
Read compact client-level AI citation and mention rates with per-platform coverage and explicit project counts. | Read-only | Free |
get_tracker_citation_analysis |
Use this when the user asks which pages AI answers or Google results cite for a tracked topic, or where a client is cited, unlinked or missing. Read the saved page-level analysis of a Tracker project’s cited (ai_citation) or ranking (traditional_search) URLs: source type, URL type, mention status, and opportunity status, matching the app’s citation tables. Paginated; reads saved data only. Do not use for an overall visibility score (use get_visibility_summary). | Read-only | Free |
get_tracker_competitors |
Use this when the user asks who outranks a client in Google for a tracked topic. Read the domains that most often rank in Google for one Tracker topic, with appearances, latest, best, and average position over the window, matching the app’s organic competitor chart. Optional per-scan history. Do not use for competitors named in AI answers (use get_visibility_summary). | Read-only | Free |
get_tracker_matrix |
Return a bounded client keyword × platform matrix with project_id, keyword, position, mention, citation, measurement state, comparison state, URLs, and freshness without one detail call per keyword. view=summary returns per-project counts plus missing and failed platform lists. client_missing_only keeps only completed checks where the client is absent. | Read-only | Free |
get_tracker_project |
Get a Tracker project with compact summary output by default. Includes explicit data state, next scheduled run, SPI, platform measurements, and comparison availability; request full only for complete provenance. | Read-only | Free |
get_tracker_results |
Read one Tracker project’s latest-per-platform, latest-run, or exact-run measurements. Every payload names project_id, client_id, and keyword. Views: summary (one line per platform), compact (default), full (complete AI answers, citations, sentiments, provider diagnostics, and saved provider-returned research_trace fan-out queries when available). Local Pack grids collapse to the representative cell with grid coverage unless local_grid is points or all. Set include_all_results=true for the full ranked lists (Google, Bing, DuckDuckGo, Brave) and Local Pack listings with rating, reviews, phone, and address. client_missing_only keeps only completed checks where the client is absent; failed checks (data_state failed, for example an empty results list) are never counted as missing. | Read-only | Free |
get_tracker_seo_performance |
Read the connected first-party SEO performance dashboard for one client over an allowed 7, 28, or 90-day window. | Read-only | Free |
get_tracker_sources |
List the pages behind one Tracker scan: every ranking URL (traditional_search) and every AI-cited URL (ai_citation), with platform, position, and whether each page mentions the client (mention_status none, linked, linked_elsewhere, unlinked, sponsored, error, or pending). Use client_missing_only=true to find analyzed pages that rank or are cited without mentioning the client organically. Sponsored placements remain separate from organic mentions. Paginated; the first page also returns a domain rollup. Reads saved Tracker data only. | Read-only | Free |
get_tracker_summary |
High-level latest-per-platform Tracker summary with explicit data_state, nullable SPI/change when unavailable, active project counts, platform coverage, and top movers. | Read-only | Free |
get_tracker_trends |
Read up to 90 days of completed or partial Tracker runs with SPI and compact per-platform measurements. | Read-only | Free |
get_visibility_summary |
Use this when the user asks how visible a client is in AI answers or search, e.g. “Is ChatGPT mentioning us?”, “How often do AI answers cite our site?” or “Where do we rank?” Read one client’s AI and local visibility across its Tracker projects from existing results (no scan, no usage): AI mention and citation rates overall, per platform, and per project; organic positions per search engine; the client’s local pack position from listing order; top competitors in AI answers and the local pack top 3; and the most-cited domains. Brand matching uses the brand name, aliases, and GBP location name, and a citation is any URL on the client’s domain. Failed or unscanned cells are excluded from rates. Compact by default; request view=full for per-cell evidence. Do not use to run a fresh scan (use trigger_scan) or to audit one page (use page_audit_run). | Read-only | Free |
list_reporter_projects |
Legacy compatibility alias for list_tracker_projects. Reporter is not the product name; use the Tracker tool ID for new workflows. | Read-only | Free |
list_tracker_projects |
List Tracker projects with compact output by default. Responses explicitly distinguish measurement and comparison states; use fields or full view only when needed. | Read-only | Free |
search_intelligence_get |
Read one organization-owned Search Intelligence run. Summary omits answer, citations, organic results, and raw payloads; full includes normalized results, and raw upstream payloads remain opt-in. Rankability Agent API reference: https://help.rankability.com/api/api-getting-started. | Read-only | Free |
search_intelligence_run |
Use this when the user asks whether ChatGPT, Google AI Mode, Perplexity, Gemini, Claude, Grok, Google or Bing mention or cite a brand or domain for a query right now, e.g. “Does ChatGPT recommend us for best CRM for agencies?” Works for any brand or domain with no client or Tracker project; pass extract_brands and extract_domains to detect them. Run one synchronous query across selected Google/AI search providers. Estimate search_intelligence_query with the provider count, show the usage impact, and obtain approval. Summary view omits answers, citations, and raw provider payloads; fetch full by run ID when needed. Do not use for ongoing tracking over time (use a Tracker project and get_visibility_summary). | Changes or deletes data | Per call |
tracker_capabilities |
Preflight the complete Tracker platform catalog, current-plan entitlement, report capacity (the organization’s report limit, current count, and remaining slots, from the same check create and upsert enforce), and client-specific YouTube, TikTok, and Local Pack dependencies before any write or usage-consuming scan. Check remaining_slots before creating a batch of reports. | Read-only | Free |
trigger_scan |
Use this when the user wants fresh AI visibility or rank tracking results for a Tracker project now instead of waiting for the next scheduled scan. Trigger a manual Tracker scan. Call estimate_usage_impact with operation trigger_scan and this project_id, show the pooled-usage impact, and obtain approval before setting confirm_usage=true. Runs asynchronously. Do not use just to read existing results (use get_visibility_summary). | Changes or deletes data | Per call |
update_tracker_project |
Update an existing Tracker report’s configuration or recurring schedule without running an immediate scan. Requires explicit confirmation and a stable idempotency key. | Changes or deletes data | Per call |
upsert_tracker_project |
Declaratively create or update exactly one client/topic Tracker report in place. Resolves plan and identity dependencies before writing, preserves existing history, supports a no-write dry run, and never starts a scan. | Changes or deletes data | Per call |
Content
Section titled “Content”| Tool | What it does | Access | Billing |
|---|---|---|---|
approve_brief |
Approve a reviewed stepped-mode brief and start draft generation. Call estimate_usage_impact for create_content_stepped, show the pooled-usage impact, and obtain approval before setting confirm_usage=true. | Changes or deletes data | Per call |
create_content_auto |
Create a new content generation job that runs the full pipeline in one call. Call get_usage_and_limits and estimate_usage_impact, show the current pooled-usage impact, and obtain approval before setting confirm_usage=true. | Changes or deletes data | Per call |
create_content_autopilot |
Start a Copywriter Autopilot run. Call get_usage_and_limits and estimate_usage_impact, show the current pooled-usage impact, and obtain approval before setting confirm_usage=true. | Changes or deletes data | Per call |
create_content_stepped |
Create a content brief for later writing. For pure entity extraction or entity research use extract_entities instead; it needs no content brief or writing approval. Review this brief with get_content_artifacts, then approve_brief to write a draft. create_content_auto runs the full content pipeline. | Changes or deletes data | Per call |
delete_content_project |
Permanently delete one Copywriter project and its artifacts. Requires the exact current topic as a second confirmation. | Changes or deletes data | Free |
estimate_entity_research |
Use before pure entity extraction for a keyword, topic or subtopic. Check the existing Quick research usage contract for 1–5 topics without starting work. Show this and obtain approval before extract_entities. Each topic creates one saved entity report. | Read-only | Free |
extract_entities |
Use for pure entity extraction, entity research, or relevant entities for 1–5 keywords/topics/subtopics. Research selected search services through the shared entity engine without creating a content brief, outline or draft. Return saved async job IDs; poll get_entity_research for compact prioritized entities. First show estimate_entity_research and obtain usage approval. | Changes or deletes data | Per call |
get_autopilot_status |
Get the live status of a Copywriter Autopilot run — current stage (queued, researching, briefing, drafting, optimizing, fact_checking, image, finalizing, complete, failed), optimize passes, whether the account entitlement was consumed, and the final review summary. Poll until stage is complete or failed, then use get_content_artifacts to fetch the finished draft. | Read-only | Free |
get_content_artifacts |
Retrieve saved draft, outline, SEO assets, word_count_target, keywords checklist, entities, FAQs and quality score. Before drafting, read word_count_target (available at brief_ready). keywords returns editor recommendations with covered/usage_count/ignored/score_eligible flags for the saved draft; page using keyword_limit/offset until has_more=false. Coverage is null before drafting; unsaved editor changes are not included. quality_score includes editor_breakdown for coverage and placement; categories_unavailable refers only to the separate seven-category model. Entities are compact and paginated by default; entity_view=full includes source excerpts. For new entity-only research use extract_entities. Request draft with quality_score; freshness_state confirms it scores that saved draft. | Read-only | Free |
get_content_project |
Get a Copywriter project’s details, lifecycle, and quality score with editor coverage/placement breakdown. target_word_count_min/max is the editor’s actual target range, available at brief_ready; use it before drafting rather than the brief’s single-number hint. A saved web draft is draft/not_started, not queued. | Read-only | Free |
get_copywriter_queue_status |
Read bounded organization-specific Copywriter research and draft queue depth, position, estimated wait, and service alerts. | Read-only | Free |
get_entity_research |
Poll an entity-only job and return a compact, paginated entity list with relevance, aliases, writing eligibility and source counts. Use view=full for literal source evidence. Follow entity_pagination.has_more to retrieve remaining entities. No new work, brief or draft. | Read-only | Free |
list_content_projects |
List user-created Copywriter projects with optional client and status filtering. Internal Page Auditor editor handoffs are excluded; use page_audit_list for audits. status=draft means a saved web-app project that has not started generation; queued is reserved for an Agent API handoff. creation_source and nullable mode expose that provenance explicitly. | Read-only | Free |
optimization_get |
Poll or retrieve one durable Optimize run. Use status while queued/running, summary for compact scores, and full only after completion when the complete result is needed. Rankability Agent API reference: https://help.rankability.com/api/api-getting-started. | Read-only | Free |
optimization_list |
List durable Optimize history for the active organization. Results are compact, bounded, and can be filtered by attributed client, status, keyword, or URL. Rankability Agent API reference: https://help.rankability.com/api/api-getting-started. | Read-only | Free |
optimize_page |
Use this when the user wants content optimization: how a page’s content scores against the pages ranking for its keyword and what to add. Start a durable asynchronous page score against SERP competitors. Call estimate_usage_impact, show the pooled-usage impact, and obtain approval before setting confirm_usage=true. Returns an optimization_id immediately; poll optimization_get. Do not use for technical or crawlability checks (use page_audit_run). | Changes or deletes data | Per call |
replace_content_draft |
Replace the complete draft body and reset scores/review artifacts tied to the old content. Requires the exact current project topic as a second confirmation and refuses active jobs. | Changes or deletes data | Free |
update_content_brief |
Patch a stepped-mode brief before approval. Omitted sections are preserved; approval and paid draft generation remain a separate action. | Changes or deletes data | Free |
Keyword research
Section titled “Keyword research”| Tool | What it does | Access | Billing |
|---|---|---|---|
get_researcher_project |
Get one organization-scoped saved Researcher project with a bounded page of its keyword set. This reads existing data and starts no research. | Read-only | Free |
list_researcher_projects |
List bounded, compact saved Researcher projects for the organization, optionally scoped to one client. Saved projects are separate from asynchronous job results. | Read-only | Free |
researcher_get |
Get one organization-scoped Researcher job. Use the bounded status view while polling; request full only after completion to retrieve the result payload. | Read-only | Free |
researcher_list |
List bounded, compact Researcher job summaries for the organization, optionally scoped to one client. | Read-only | Free |
researcher_run |
Use this when the user wants keyword research, topic ideas or content opportunities, e.g. “What should we write about next?” Works without a client; pass client_id to ground it in that client’s Knowledge and Search Console data. Start an asynchronous Researcher keyword-research job. Cache hits may complete immediately without usage. Otherwise call estimate_usage_impact for researcher_run, show the current windows, obtain approval, and set confirm_usage=true. Poll researcher_get with view=status, then request view=full after completion. Do not use to check current rankings (use get_visibility_summary or get_gsc_search_performance). | Changes or deletes data | Per call |
Audits, crawling and extraction
Section titled “Audits, crawling and extraction”| Tool | What it does | Access | Billing |
|---|---|---|---|
audit_manifest_get |
Return a canonical client artifact manifest across Tracker, Page Auditor, Site Auditor, Researcher, and Copywriter with lifecycle, supersession, retry provenance, configuration hashes, and batch relationships. | Read-only | Free |
automation_audit_handoff |
Read a completed automation audit’s technical-only findings and reusable agent prompt. Follow every findings page. Content recommendations and unknown issue types are excluded. Includes unavailable checks, saved coverage and draft-only/no-publish restrictions. Requires routines:read and site-auditor:read. Use the member’s authorized repository/CMS access and verify evidence before drafting fixes; never merge or publish automatically. | Read-only | Plan only |
automation_audit_run |
Run one audit with the saved automation scope, preserving a new canonical report. Show site_audit_estimate or estimate_usage_impact for saved maxPages and obtain approval first. Requires routines:write and site-auditor:write. Weekly scheduling is unchanged. Reuse the idempotency key for transport retries. No external agent is invoked and no fixes are applied. | Creates or updates | Plan only |
batch_url_get |
Read a bounded page of one saved Batch URL Analyzer result. Null measurements remain null and each row is labeled complete, partial, or error. This starts no analysis. | Read-only | Free |
batch_url_list |
List bounded saved Batch URL Analyzer run metadata for one client. This does not analyze URLs, call providers, or expose legacy credit units. | Read-only | Free |
extract_page_data |
Extract a grounded answer, highlights, or schema from exactly one URL, organization-owned crawler page, or supplied text. Estimate crawler_extract and obtain approval first. Missing evidence remains null or explicitly unanswered. Rankability Agent API reference: https://help.rankability.com/api/api-getting-started. | Changes or deletes data | Per call |
gbp_audit_get |
Read one saved GBP audit. Summary omits raw review and question bodies; full adds bounded checkpoints and competitor aggregates but still omits those bodies. No provider is called. | Read-only | Free |
gbp_audit_list |
List bounded saved Google Business Profile audit artifacts for one client. This read never resolves a listing, calls Google/SerpAPI, starts an audit, or changes usage. | Read-only | Free |
get_crawl |
Poll one organization-owned crawl. Status omits pages; summary adds fetch/failure totals; full returns a paginated, field-selected page inventory with compact fields by default. Rankability Agent API reference: https://help.rankability.com/api/api-getting-started. | Read-only | Free |
get_crawler_usage |
Read the current organization daily crawler page quota for this instance without fetching a page or starting work. | Read-only | Free |
page_audit_batch_upsert |
Preflight and start up to 25 Page Auditor audits as one durable batch. Use dry_run first; retries with the same key resume the same batch and never create duplicate audits. Rankability Agent API reference: https://help.rankability.com/api/api-getting-started. | Changes or deletes data | Per call |
page_audit_delete |
Permanently delete a finished Page Auditor audit (status ‘complete’ or ‘error’). Refuses to delete an audit while it is still in progress (any other status) — wait for it to finish, then delete. This cannot be undone. | Changes or deletes data | Free |
page_audit_get |
Fetch a Page Auditor audit by ID. Use the bounded status view while polling, then full for scores, gates, technical health, agent readiness, and quality analysis. Nested output uses canonical snake_case. | Read-only | Free |
page_audit_list |
List Page Auditor audits with optional filtering by client and status. Compact is the bounded default; request full or explicit fields only when needed. Sorted by most recent first. | Read-only | Free |
page_audit_run |
Use this when the user wants to know why a page is not ranking or how to improve it for a keyword, e.g. “Audit this URL for ‘best CRM’.” Works on any public URL; a client is optional. Run a Page Auditor audit for a public URL against a target keyword. Call estimate_usage_impact, show the pooled-usage impact, and obtain approval before setting confirm_usage=true. Returns immediately with an audit_id; poll page_audit_get until terminal. Do not use for a full-site technical crawl (use site_audit_run). | Changes or deletes data | Per call |
scrape_page |
Fetch and structure one public page. Check crawler quota and estimate crawler_scrape, show the impact, obtain approval, then confirm. A query returns at most 8,000 characters of relevant content; an eligible freshness-cache hit avoids a new fetch. | Changes or deletes data | Per call |
site_audit_cancel |
Abort an in-progress Site Auditor crawl and flip the project status to cancelled. Only works while the crawl is running. Returns the updated project. | Changes or deletes data | Free |
site_audit_delete |
Permanently delete a finished Site Auditor project and all of its crawled pages + detected issues. Refuses to delete a project while its crawl is in progress (‘crawling’ or ‘analyzing’) — cancel the crawl first with site_audit_cancel, then delete. This cannot be undone. | Changes or deletes data | Free |
site_audit_estimate |
Check the pooled-usage impact of a proposed Site Auditor crawl without starting work. Larger page ceilings may have high impact; the response contains no credits or per-action price. | Read-only | Free |
site_audit_get |
Fetch a Site Auditor project by ID without forcing its full inventory into every call. Use status while polling, summary for bounded findings and verification totals, or full with page/issue pagination and explicit page fields. Full inventory omits page bodies by default; use site_audit_get_page when body evidence is needed. | Read-only | Free |
site_audit_get_page |
Read one page from a Site Auditor project. Summary is compact; content adds the complete analyzed page text and duplicate passages; full returns every supported stored page field. Use explicit fields for the smallest response. | Read-only | Free |
site_audit_list |
List Site Auditor projects with optional filtering by client and status. Sorted by most recent first. | Read-only | Free |
site_audit_run |
Use this when the user wants a technical SEO audit of a whole website: broken links, indexability, crawl issues. Create a Site Auditor project and start a crawl. Call site_audit_estimate, show the pooled-usage impact, and obtain approval before setting confirm_usage=true. Returns immediately with a project_id; poll site_audit_get with view=status until terminal. Do not use to evaluate one page against a keyword (use page_audit_run). | Changes or deletes data | Per call |
start_crawl |
Start an asynchronous same-site crawl of at most 200 pages and depth 5. Read crawler quota and estimate crawler_crawl with the intended page ceiling before approval; poll get_crawl with view=status. | Changes or deletes data | Per call |
Links and outreach
Section titled “Links and outreach”| Tool | What it does | Access | Billing |
|---|---|---|---|
backlink_profile_refresh |
Create or refresh the client’s four saved backlink-profile sections. Call estimate_usage_impact for backlink_profile_refresh, show the impact, obtain approval, then confirm. Fresh snapshots are safely reused for six hours. | Changes or deletes data | Per call |
get_backlink_profile |
Read one bounded section of the client’s saved backlink-profile evidence with its cache timestamp and freshness state. This never invokes DataForSEO, refreshes data, loads individual live backlinks, exports rows, or consumes usage. | Read-only | Free |
prospector_add_item |
Save one known public URL to a Prospector outreach list. Canonical URLs are de-duplicated within the list. This does not fetch, discover, analyze, enrich, email, or contact the prospect. Rankability Agent API reference: https://help.rankability.com/api/api-getting-started. | Creates or updates | Plan only |
prospector_create_list |
Create one saved outreach list for a client. This creates workspace state only; it does not discover or enrich prospects. | Creates or updates | Plan only |
prospector_delete_item |
Permanently remove one saved item from a Prospector list. Requires the exact current URL as a second confirmation; it does not delete the source page or contact record outside Rankability. | Changes or deletes data | Plan only |
prospector_delete_list |
Permanently delete one saved Prospector list and its list memberships. Requires the exact current list name as a second confirmation. | Changes or deletes data | Plan only |
prospector_list_items |
Read a bounded page of saved outreach items from one Prospector list. Filter by pipeline state without loading discovery or enrichment payloads. | Read-only | Plan only |
prospector_list_lists |
List bounded saved outreach lists for one client, including item counts and pipeline-status breakdowns. This reads saved workspace data and does not run discovery, enrichment, or backlink providers. | Read-only | Plan only |
prospector_update_item |
Update one saved outreach item’s pipeline status or notes. This records workflow state only and never sends outreach. | Changes or deletes data | Plan only |
Publishing, routines and automations
Section titled “Publishing, routines and automations”| Tool | What it does | Access | Billing |
|---|---|---|---|
automation_get |
Read current Automate settings/history before changing them. Content results omit topic bodies; use routine_get for one schedule’s paginated topics if needed. Follow pagination.hasMore using offset. Technical audits use version as expectedVersion and clientDomain as expectedDomain when updating. Requires routines:read; no work is started. | Read-only | Plan only |
automation_list |
Read all Automate workflows for a client: technical audits, content, Knowledge monitoring, brand mentions, Google and Yelp reviews. Includes settings and bounded history; unavailable workflows are explicit. Requires routines:read. Never runs or changes anything. | Read-only | Plan only |
automation_update |
Create, enable, pause or change one Automate workflow through its canonical settings handler. Read automation_get first and preserve settings the user did not request changing. Enabling authorizes recurring allowance/usage, and existing content/review workflows may publish according to their chosen mode. Show exact settings and cadence for approval. Technical audits stay technical-only, weekly by default, 100 pages by default, UTC scheduling, with manual agent handoff and draft-only fixes. Requires routines:write. Does not run an immediate audit. | Creates or updates | Plan only |
publishing_get_delivery |
Read one privacy-safe publishing receipt with preflight, remote identity, source-currency, and validation state. Credentials and article bodies are never returned. | Read-only | Plan only |
publishing_list_connections |
List credential-free publishing connection status, safe target identity, capabilities, and requirements for one client. This never tests, reconnects, or changes a destination. | Read-only | Plan only |
publishing_list_deliveries |
List bounded privacy-safe delivery receipts for one client. This read never retries, validates, creates, updates, or publishes remote content. | Read-only | Plan only |
publishing_preflight |
Check one immutable article snapshot against a connected destination’s shared mapping and readiness rules. This read-only check never creates a delivery, calls the destination, spends credits, or grants live-publishing approval. | Read-only | Plan only |
publishing_prepare_artifact |
Create or reuse an immutable, destination-specific snapshot of a saved Copywriter draft. This internal preparation step does not create a CMS item, open a GitHub pull request, contact the destination, or approve publishing. | Creates or updates | Plan only |
publishing_request_draft_delivery |
After publishing_preflight is ready and the user explicitly approves the reviewed destination and article snapshot, create an idempotent destination draft or reviewable GitHub pull request. Live publishing is not supported. GitHub requires the exact preview hash returned by preflight. | Changes or deletes data | Plan only |
routine_get |
Read one recurring content or Knowledge-monitor Routine. Content topics or monitored source IDs are paginated; suggested/published topic bodies remain omitted. | Read-only | Plan only |
routine_list |
List bounded recurring content and Knowledge-monitor Routine state for one client. This read never enables, schedules, runs, or changes a Routine. | Read-only | Plan only |
Campaigns
Section titled “Campaigns”| Tool | What it does | Access | Billing |
|---|---|---|---|
append_campaign_log |
Append an action, decision, note, or handoff to the campaign log. Decisions start as proposed. To approve or reject a proposed decision, append a decision with resolves_entry_id and decision_status approved or rejected; to close a handoff, append a note or action with resolves_entry_id. History is never rewritten. | Creates or updates | Plan only |
attach_campaign_document |
Attach an existing client Knowledge source to the campaign as an internal or client-facing document. Re-attaching the same source updates its audience and note. To add new text, create the Knowledge source first with create_client_source. | Changes or deletes data | Plan only |
compare_campaign_snapshots |
Compare two saved campaign snapshots (defaults: the first baseline against the latest). Returns headline changes marked improved, declined, unchanged, or not comparable, the Tracker visibility comparison when recorded, and notes when the prompt set or date ranges differ. | Read-only | Plan only |
create_campaign |
Create the campaign record that becomes the source of truth for a bounded client campaign (goal, market, dates, context, and a prompt set that references existing Tracker reports). This does not create, schedule, or scan Tracker reports. | Creates or updates | Plan only |
get_campaign |
Compact campaign summary: goal, market, dates and week N of M, context facts/rules/open questions, prompt set with each Tracker report’s latest persisted scan status, latest snapshot headline numbers, the last N log entries, open decisions and handoffs, and attached Knowledge documents. Log detail text is omitted unless include_log_detail=true. | Read-only | Plan only |
list_campaigns |
List a client’s campaign records (name, goal, market, status, dates, week N of M). Archived campaigns are hidden unless requested. | Read-only | Plan only |
save_campaign_snapshot |
Create and persist a labeled campaign snapshot (default ‘Week 0’ first, then ‘Week N’) of the campaign’s prompt-set visibility from stored Tracker results, plus the Search Console clicks/impressions and GA4 organic sessions/key events the SEO performance dashboard shows for a date range. Each first-party source records its coverage (days with data vs days in range) and a partial flag; treat a partial total as incomplete, not as a baseline. This writes a new snapshot record from existing data; it never starts a scan or a sync and uses no credits or usage. Retrying with the same idempotency key returns the saved snapshot without creating another record. | Creates or updates | Plan only |
update_campaign |
Patch a campaign’s name, goal, market, category, dates, status, context lists, or prompt set. Omitted fields are preserved; a provided context list replaces the stored list. Set prompt_set to null to unlink the prompt set. Tracker reports themselves are never changed. | Changes or deletes data | Plan only |
Account, usage and help
Section titled “Account, usage and help”| Tool | What it does | Access | Billing |
|---|---|---|---|
estimate_usage_impact |
Check whether an on-demand operation has standard or high pooled-usage impact without starting work. Use this before approval. The response contains no credits, currency, or per-action price. | Read-only | Free |
exports_get |
Retrieve a browser-free JSON or CSV artifact manifest or paginated result export. Result rows cover Page Auditor, Site Auditor pages, or Tracker runs with target/ranking URLs, platform position, mentions, citations, scores, source URLs, and run dates where applicable. | Read-only | Free |
get_notifications |
List recent alerts and anomalies: GSC traffic anomalies and citation alerts. | Read-only | Free |
get_usage_and_limits |
Check the pooled general and burst usage windows before on-demand work. The response includes the connected organization (id and name), remaining percentages, recovery times, protected scheduled work, and the billing usage-page link. This read starts no work and consumes no usage. | Read-only | Free |
get_usage_events |
Read a bounded, paginated history of completed usage outcomes for the active organization. Shared-credit responses include the signed credit amount and resulting balance. Pooled-usage responses identify included work without exposing an invented per-action price or internal capacity units. | Read-only | Free |
jobs_status |
Reconcile one or up to 50 durable Rankability job IDs across Page Auditor batches and audits, Optimize, Researcher, Tracker scans, and Site Auditor, or pass client_id to read every Tracker scan for that client in one call. The client view reports each project’s scan_state (queued, running, complete, partial, failed_needs_rerun, canceled, never_scanned), needs_rerun, the latest run with failed platforms, and the last successful run. Returns polling guidance. | Read-only | Free |
search_help_center |
Search the canonical Rankability Help Center with bounded results. Compact view returns titles, summaries, categories, and canonical links; full includes article Markdown. | Read-only | Free |
