Skip to main content
The Hikoo MCP currently ships 20 tools across 8 areas. You never call them directly. Describe what you want in natural language and your AI client picks the right one. All tools operate on the single website bound to your API key. No website_id parameter is needed.

​
Website

​
hk_get_website
read-only
Returns the website metadata, topics, competitors list, language, and subscription status.

​
Audits

​
hk_list_audits
read-only
List every audit ever run for this website (most recent first), with scores, status, and audited-page count.
​
hk_get_audit
read-only
Detail of a single audit: overall score, timing, and the list of audited pages.Args: audit_id
​
hk_get_audit_page
read-only
Page-level audit result with every criterion and its validation status, plus the recommendations attached to each failing criterion. This is the most detailed audit view.Args: audit_id, page_id
​
hk_start_audit
write
Queue a new full-site audit. Runs asynchronously. Poll hk_list_audits or hk_get_audit for status.Fails if the website’s audit quota for the current billing period is exhausted.

​
Citations

​
hk_list_citations
read-only
Paginated list of raw AI citations: the actual content the LLM produced, its sources, and which competitors were mentioned.Args: page, page_size, time_range, website_cited (filter to cited/not-cited only)
​
hk_get_top_platforms
read-only
Top AI platforms by citation count.Args: time_range
​
hk_get_platforms_share_of_voice
read-only
Visibility trend data per platform (share of voice over time).Args: time_range
​
hk_get_top_sources
read-only
Top source domains (or URLs) that appear in AI citations. Useful to understand where LLMs get their answers.Args: time_range, group_by (domain or url), limit

​
Prompts

​
hk_get_prompts_quota
read-only
Current prompt count vs the subscription limit. Call this before bulk-creating prompts. Every create past the limit fails.Returns { used, limit, remaining }.
​
hk_list_prompts
read-only
All prompts tracked for the website, grouped by topic, each with its visibility score.Args: topic_id (optional filter)
​
hk_create_prompt
write
Create a new prompt. Fails if the website’s prompt quota is reached. Check hk_get_prompts_quota first.Args: content, topic_id, localizations
​
hk_update_prompt
write
Update an existing prompt. Only fields you provide are changed.Args: prompt_id, content, topic_id, localizations
​
hk_delete_prompt
destructive
Delete a prompt.Args: prompt_id
​
hk_get_prompt_citations
read-only
Paginated citations tied to a specific prompt.Args: prompt_id, page, page_size

​
Topics

​
hk_list_topics
read-only
List every topic tracked for the website.

​
Competitors

​
hk_list_competitors
read-only
Competitor ranking (share of voice) against your website.Args: time_range

​
Source Intelligence

Discover and prioritize the AI-cited domains around your space — review platforms, directories, blogs, media — so an agent can automate listing claims, guest-post pitches, and review outreach.
​
hk_get_source_intelligence_stats
read-only
Aggregated counts: total sources, breakdown by source_type and by AI platform. Call this first to know what kinds of sources exist before listing.Returns { total, by_type, by_platform }.
​
hk_list_suggested_sources
read-only
Paginated list of suggested sources, ordered by citation_count desc. Filter by type to pick the right outreach motion (review listings vs guest posts vs PR).Args: source_type, platform, search, limit, offsetsource_type values: review_platform, media, blog, directory, forum, wiki, social. platform values: chatgpt, perplexity, gemini, claude — only sources cited by that platform.
Visibility is capped to your plan’s top-N most-cited sources. Upgrade unlocks more sources, not different data.

​
Health

​
hk_get_health
read-only
Global website health (AI crawlers allowed, llms.txt present, sitemap reachable) plus actionable improvement suggestions.Args: include_dismissed

​
Argument conventions

  • time_range accepts: all_time, 1y, 1m, 1w, 1d.
  • page / page_size are 1-based. page_size is capped at 100.
  • Write tools mutate data in your workspace. Claude Desktop and Claude Code ask for confirmation before running them by default.

​
Annotations

Each tool is labelled so MCP clients know how to handle it:
LabelMeaning
read-onlyNever mutates data. Safe to call freely.
writeCreates or updates data.
destructivePermanently removes data. Always review before confirming.