The GetIntel MCP server exposes 42 tools: 30 that read your data and 12 that act. The action tools work on your task list, the technical scan, pitch reports, and the topics and prompts you track. Every tool works on the brand the token was created for.
Tools marked period default to the last 7 days. Pass one of these to change it; the response’s period says what was used.
| Input | Type | Meaning |
|---|
days | integer | Days back, ending today. 1 to 90. Ignored when from is given. |
from | string | Start date, YYYY-MM-DD. At most 90 days before to. |
to | string | End date, YYYY-MM-DD. Defaults to today. |
A brand on Pro or Growth can be tracked in more than one market: a country plus a language, each with its own questions (asked in that language), answers, scores, competitors, sources and tasks. Every tool except get_usage and list_markets takes an optional market input:
| Input | Type | Meaning |
|---|
market | string | A market code from list_markets, such as IN-hi (or just IN), or the market’s id. Defaults to the primary market. |
Brand perception (get_brand, get_brand_transcript) and the technical scan are the same in every market.
| Tool | What it returns | Inputs |
|---|
get_overview | Everything on the Home page: your visibility, share of voice, position and citations against your top rivals (for the period and per day), how AI perceives you, which of your pages AI cites, topics weakest first, AI crawler and visitor totals, the technical readiness score, and open tasks per layer. | period |
list_markets | The markets the brand is tracked in: code, country, language, whether it’s the primary market, and how many topics and questions it tracks. Does not count against your quota. | none |
get_usage | This month’s MCP reads used, the limit, what remains and when it resets. Does not count against your quota. | none |
| Tool | What it returns | Inputs |
|---|
get_topics | Every topic you track: your share of AI answers in it, the change over the period, its rank, and the rivals AI names instead. | period |
get_topic | One topic in full: every buyer question with each engine’s latest answer, the head-to-heads against rivals, what AI cites for the topic, the score’s history with tracked_set_changes (the days its questions changed), and the content analysis (the facts AI compares, and what a page needs to say to get cited). | topic_id |
get_visibility | The brand-wide visibility score (0 to 100) across all tracked questions for the period, the change against the period before, and 12 weeks of history. Also returns tracked_set_changes: the days in the period that topics or prompts were added, archived, reworded or moved, with the prompt count after each. A null score means nothing was measured; 0 means AI answered and named you nowhere. | period |
get_prompt_history | Which buyer questions you won, lost or held between this 7-day window and the one before, with the mention rate in each. | none |
| Tool | What it returns | Inputs |
|---|
get_ai_answers | Each buyer question with the rate you were named over the last 7 days and each engine’s state: recommended, mentioned, absent, open (no product named at all), no_answer or error. | none |
get_prompt_matrix | Question × engine grid: each engine’s most common state over 7 days, the rate behind it, and your position when named. | none |
get_receipts | The exact answer each engine gave to one buyer question over the last 7 days. Returns answer_rate (share of individual answers naming you) and scan_rate (share of scans where at least one engine named you, the rate get_ai_answers reports). | prompt (text from get_ai_answers) |
| Tool | What it returns | Inputs |
|---|
get_competitors | Share of voice: how often each engine names you versus each competitor, and on which questions. our_visibility_rate matches the app’s Home and Prompts pages for the same period. | period |
get_competitor | One rival in depth: the topics they beat you on and by how much, the questions they win that you lose, what AI said, and their pages AI cites. | name (as AI writes it) or competitor_id |
| Tool | What it returns | Inputs |
|---|
get_brand | Whether AI knows your brand and would recommend it, per engine (ChatGPT, Perplexity, Gemini, Google AI Overviews): whether it describes you accurately, whether you come up unprompted in a category question, and its reasoning on recommending you. Weekly runs, with the trend over the period. | period |
get_brand_transcript | The full prompts, answers and citations behind get_brand. | engine (optional) |
| Tool | What it returns | Inputs |
|---|
get_sources | The domains AI cites in your category, how often, and whether each is yours, a competitor’s or third-party. | period, limit (1–50, default 15) |
get_your_pages | Which of your pages AI cites, now versus the previous period, by engine and topic, the questions each is cited for, and what AI cites instead of your uncited topic pages. | period |
get_content_gaps | For each buyer question: whether you have a page, whether AI cites it, and the recommended action (write, improve or leave). | none |
| Tool | What it returns | Inputs |
|---|
get_technical | The latest weekly technical scan: an AI-readiness score per engine, every check by layer, each key page’s issues, bot access, and what is fixed or new since the last scan. | none |
get_ai_crawlers | How often ChatGPT, Perplexity, Claude and Gemini bots read your pages, verified against each vendor’s published IPs, with pages newly cited or cited less. Needs Cloudflare or the GetIntel server script. | period |
get_ai_visitors | People who arrived from an AI assistant, by assistant and landing page, versus the previous period. Needs Google Analytics 4 or the GetIntel visitor tag. | period |
get_search_console | Search Console clicks and impressions joined to your tracked questions, including pages and queries that rank in Google’s top 10 but that AI never cites. Needs Search Console connected. | days (1–90, default 28), dimension (query or page), limit (1–100, default 25) |
| Tool | What it returns | Inputs |
|---|
get_signals | What buyers ask on Reddit, grouped by topic: the threads (and whether AI cites them) and each topic’s weekly summary. | period, limit (threads per topic, 1–50, default 20) |
| Tool | Access | What it does | Inputs |
|---|
list_tasks | Read | Your tasks, best first, with open counts per layer. Filter by layer (on_page, off_page, ugc, technical) or status (open by default, in_progress, done in the last 30 days, snoozed). | layer, status, limit (1–100, default 30), offset (page with next_offset; total gives the full count) |
get_task | Read | One task in full: why it matters, what to do, the topics it moves, and whether it is on this week’s plan. | task_id |
get_weekly_plan | Read | This week’s plan and a few top suggestions to add. | none |
update_task | Action | Change one task like the buttons on its card: set status (open, in_progress, done, dismissed), add or remove it from this week’s plan with planned, or hide it with snooze_days (0 to unsnooze). | task_id, status, planned, snooze_days |
plan_week | Action | Fill this week’s plan with the top tasks per layer: 2 on-page, 2 off-page, 1 UGC, 1 technical. refresh: true re-picks tasks not yet started. | refresh |
recheck_technical | Action | Run a technical scan now instead of waiting for the weekly one. At most once an hour; results appear in get_technical a few minutes later. | none |
Change what GetIntel measures: the same + Add topic, Add prompt, Archive and Restore actions as the Topics and Prompts pages, with the same limits. All of them need action access.
| Tool | What it does | Inputs |
|---|
add_topic | Start tracking a topic in this market. GetIntel writes its buyer questions from description and scans them, so it uses questions from your limit. If an archived topic has the same name, the error tells the agent to restore it instead. | name, description (the buyer situation, at least 20 characters), category (default use_case), priority (1–3, default 2) |
archive_topic | Stop tracking a topic in every market. Its questions stop being scanned and free their slots, and open tasks for only this topic are put away. History is kept. Two steps: see below. | topic_id, confirm |
restore_topic | Track an archived topic again, in the markets it was in, with fresh questions. An unknown id lists the brand’s archived topics. | topic_id |
add_prompt | Track one more buyer question in this market and scan it straight away. Pass the topic it belongs to so it counts in that topic’s score; it’s required once the brand tracks topics. | prompt, topic_id |
archive_prompt | Stop tracking one prompt. Its answers stay in history. Two steps. | prompt_id (the probe_id from get_ai_answers), confirm |
restore_prompt | Track an archived prompt again. Needs a free slot. | prompt_id |
edit_prompt | Reword a prompt. The old wording is archived with its history and the new wording is tracked under the same topic, linked to it, starting from no answers. Works when every slot is full, since it swaps one for one. Two steps. | prompt_id, prompt (the new wording), confirm |
move_prompt | File a prompt under another topic, so it counts in that topic’s score instead. The brand-wide number doesn’t change. | prompt_id, topic_id |
How these tools protect your numbers:
- Two-step archive and reword. Called without
confirm: true, archive_topic, archive_prompt and edit_prompt change nothing. They return what would happen: how many questions and tasks would go, and how many prompts visibility would then be calculated over. Your agent should show you that and call again with confirm: true only once you agree. These tools are also marked destructive, so clients that ask before risky actions will ask.
- Slots in every answer. Each reply includes
prompt_slots (used, limit, remaining) and prompts_scored, the number of prompts visibility is now calculated over. When every slot is taken, adds are refused with a message saying to archive one first.
- Daily limit. An agent can add 2 topics and 10 prompts a day per brand (restores and rewordings count). Every add starts real scans, so an agent stuck in a loop stops instead of using up your scans. The app itself has no daily limit.
- Not during the free trial. Like archiving in the app, these tools need a paid plan.
- Every change is marked. Adding, archiving, rewording or moving prompts moves your visibility by itself. Each change, from the app or an agent, is marked on the Home visibility chart and returned by
get_visibility, so a jump caused by new prompts doesn’t read as a change in AI answers.
- GetIntel’s own validation, sentiment and head-to-head questions can be archived but not reworded or moved.
Prospects you scan to win the work — companies that aren’t clients. Account-level: they answer the same whichever client you’re pointed at, and they take no market argument. See Pitch reports.
| Tool | Access | What it does | Inputs |
|---|
list_pitches | Read | Every prospect on the account with its visibility, report status, share link and how often that link was opened — plus how much of the month’s allowance is left. | limit (1–100, default 30) |
get_pitch | Read | One prospect in full: market, business type, and its report with the link to send them. | pitch_id |
create_pitch | Action | Scan a company that isn’t a client yet: reads their website, works out what they sell and to whom, and runs one scan. The report is ready a few minutes later from get_pitch. Spends one of the month’s pitch reports. | url, market_code (e.g. GB-en; default: whatever their site reads as) |
create_pitch runs a real scan against a company that isn’t a customer, so it needs action access and it isn’t idempotent — calling it twice scans twice and spends twice. When the month’s allowance is gone it says so rather than scanning anyway.
These still respond but belong to earlier parts of GetIntel. New work happens in tasks and the weekly plan.
| Tool | What it returns |
|---|
list_activity | Recent signals, approvals, drafts and outcomes, newest first. |
list_drafts / get_draft | Drafts written by the earlier mission queue. No new drafts are created there. |
- Plan required. Every tool needs an active GetIntel plan. Action tools are refused on an inactive plan with a message saying so. The topic and prompt tools also need a paid plan, and have a daily limit.
- Monthly read quota. 10,000 reads on Monitor, 50,000 on Pro and 200,000 on Growth, resetting on the 1st.
get_usage and list_markets are free to call.
- Readable errors. A missing integration (for example Search Console not connected), an unknown id or a failed third-party call comes back as an error message your agent can act on, never as an empty result.
