MCP App Store
by KeenEthics
Frontmost AI icon

Frontmost AI MCP Server for ChatGPT

By Frontmost AI

Overview

Help buyers discover your business through AI search. See when AI apps mention your business, which rivals they name, and the pages they cite. Ask your assistant to turn those answers into a draft page edit or a note for a specific publisher. Update tracked questions and share scan reports from the same conversation. An existing Frontmost AI account is required.

How to connect

  1. ChatGPT: Open the Frontmost AI listing in the ChatGPT app directory and choose Connect. ChatGPT listing Frontmost AI exposes 24 tools here.

Tools

create_project

ChatGPT
Start tracking a business in one market. Give its website domain. Frontmost AI identifies the business, infers its market, writes 12 questions (9 buyer questions and 3 that name the business, fewer when the plan's monthly credits do not cover 12) unless you pass questions, and starts the first scan with the run_config. The project must fit the plan's projects. If the result has status "drafting", call create_project again with its draft_id and the run_config it returns. Read progress with get_scan.

delete_project

ChatGPT
Delete a project with its scans, answers, and recommendations. This cannot be undone. Give the project's exact name as confirm_name, as the app's settings page asks. It and its questions stop counting toward the plan.

edit_profiles

ChatGPT
Add, rename, describe, or remove a project's profiles in one batch: the kinds of buyer it sells to, up to 4. A profile has a name and who the buyer is. Put questions in a profile with edit_questions. Removing a profile keeps its questions in the project.

edit_questions

ChatGPT
Add, pause, resume, move, or reword a project's questions in one batch, put them in a profile, or say whether one names the brand or targets a place. Copy question IDs from get_questions or get_scan, and profile IDs from get_profiles. A paused question keeps its history. New text rewords a question: the old wording pauses and the new wording joins, so each keeps its own history. The result gives the plan's questions in use and each question's is_branded (null while not sorted), classification_status, classification_source, is_location (null while not sorted), and location_source. A batch that adds questions past the plan's limit is rejected.

edit_scorecard_views

ChatGPT
Save the Scorecard's views: send every view, in tab order, as the new list. Read get_scorecard first and send its views back with the change; a view left out is deleted. Names are unique, up to 60 characters. group_by decides each group: metric, business, app, profile, or topic. items is "all" or keys from get_scorecard's items_by_group. metrics: business takes visibility, position, sentiment, win; app takes visibility, position, sentiment; profile and topic take visibility; metric takes none (its items are the metrics). break_down (default true) shows the rows under each headline: each AI app under a metric, or the other metrics under a business or AI app.

get_answers

ChatGPT
Read the saved AI answers behind a number. Copy an answers_filter from list_projects or get_scan, or filter by scan, question, AI app, named, rival, sentiment, or cited domain. Each result gives the sentences that name the business, its position, and the cited pages. A Reach page's answers_filter opens the exact answers counted for that page. page_url alone lists the answers that cite that one page, and since adds every delivered scan from that date through the scan, so a count since a page went live is exact. Pass answer_id for the full answer text and ChatGPT's web searches; it also opens an old Head to head answer. Reuse next_cursor for the next page.

get_brief

ChatGPT
Read the Briefs view. daily_brief is the latest short note after a scheduled scan. It is available on daily plans and Starter. It compares the brand's count with the previous scan. It includes a summary, questions won and lost with their AI apps, new rivals, and one recommendation to act on. daily_brief is null until a note is written. brief is the latest Weekly Deep Dive (or one earlier one by brief_id), with its week, headline, takeaways, results of past actions, findings with quotes, and three actions, each marked as from Recommendations, replacing one, or new this week. A Weekly Deep Dive arrives each Monday; this also says whether one is being written now.

get_leaderboard

ChatGPT
Read Frontmost AI's public leaderboards: which companies AI apps name most for an industry's buyer questions, from the latest weekly scan. Omit industry for the list of leaderboards. Filter by AI app or topic to rank by that slice. Pass company_id for the answers that name that company, 10 at a time.

get_profiles

ChatGPT
Read the Profiles tab: each profile's Visibility, Win rate, and Share of voice in the answers to its questions, against the top 5 rivals and on each AI app, with Visibility in each of the last 12 scans.

get_queries

ChatGPT
Read the Queries tab: the web searches ChatGPT ran while it answered, in the shown scan's week, as terms with the answers that used each one, the answers among them that named the business, change, real example searches, and the questions behind them. Also the brands that ChatGPT searched for by name. Only projects that ask ChatGPT have it.

get_questions

ChatGPT
Read the Questions tab: every question by topic, with its visibility, average position, and main sentiment, and each topic's totals. Each question has its result per AI app with the business that answer named first, won or lost since the previous scan, and the answers that named the business in each of the last 12 scans. Pass question_id for one question's page: every AI app's answer, position, the businesses and sites in those answers, its full history, and the web searches ChatGPT ran. is_location says whether a question asks about a place (location_source: auto or override; null while unknown).

get_readiness

ChatGPT
Read the AI Readiness tab: whether each AI app in scope can use the project's website, from the shown scan's check of its home page and robots.txt. status is blocked when any check fails, not_verified when none fails but one could not run, else ready. apps_ready of apps_total can use it; each app has its blocker (like robots.txt line 277), its checks, and how many of its answers cite the site. checks lists the five checks (search crawlers allowed, Gemini allowed, home page indexable, quotable, and its text readable without JavaScript) with the evidence and what to change for each one that fails. checked is false for a scan with no check.

get_rivals

ChatGPT
Read the Rivals tab: every business that AI named in the scan, with answers, named instead of you, named first, position, share of voice, change, and history. Also the top 5 per AI app and each topic against you and the top 5 rivals. head_to_head is always null: that view was retired. Pass rival_id for one rival's page.

get_scan

ChatGPT
Read one project's scan compared with the previous scan. Omit scan_id for the latest delivered scan. Set period to week, 4w, month, or quarter to add scans together. A week runs Monday to Sunday. This is the project's Overview: KPIs per AI app with win rate, the recent trend of Visibility, Position, Sentiment, and Win rate (the Key Metrics tab), the top 10 rivals (with Sentiment for the 5 tracked rivals) and cited sources, the top 5 pages by Reach, every question with its result, recommendations, the industry benchmark, and the PDF report link. A scan is usable when more than 70% of its questions are complete. metric_breakdown always gives total, branded, and unbranded metrics in the same AI app, period, and comparison, whatever questions is. Its rates are null without successful answers. Coverage gives the question counts, and issues identifies unavailable answers. Queued and running statuses mean that the scan is unfinished. Read again later. Do not start another scan.

get_scorecard

ChatGPT
Read the Scorecard tab: the project's saved views in tab order, and one view as a sheet with a row per figure and a column per period (day, 7d for trailing 7 days, week, or month), counted over every delivered scan. Each value has its counts. A headline row carries its change from the column before when the view shows change rows. break_down shows or hides the rows under each headline for this read, as the tab's Show Breakdowns switch does. Omit view for the first tab. items_by_group lists the keys that edit_scorecard_views takes.

get_sources

ChatGPT
Read the Sources tab, which adds up the shown scan and the 4 weeks before it: the 50 most cited pages to get on (cited lists, directories, review sites, and forums that do not list you, with which rivals they list); reach, the Pages by Reach list: the 50 cited pages with the widest reach in the scan's 28-day window, each with its 0 to 10 Reach and band (unbranded questions only, the same whatever questions is; Reach is the page's percentile among all of the project's cited pages in that window, by one weighted value of its citing answers (the most weight), distinct questions, AI apps, and citation share; weights gives the exponents), its cited answers, questions, AI apps, citation share, and combinations, its listing status, and get_on for Pages to Get On; and the 50 most cited sites with their share of answers, whether those answers named you, whether their pages list you, how many of the scans cited them, and history (pages_total and sites_total count them all), plus the subreddits AI cited, with their threads and whether the read threads list you. Pass domain for one site's page, with its cited pages and the businesses in those answers, or domain reddit.com with subreddit for one subreddit. Pass since to count every delivered scan from that date through the shown scan in place of the 4 weeks (Reach keeps its 28 days); counted.window states the window, so print it beside the counts. Pass page_urls to read those pages again, whatever their state now: each page's url is its row key, and a page that now lists you has left pages_to_get_on. Each site and page has its channel: own_website, competitor_sites, or the site's saved type, such as editorial or review_sites; null when the site has no channel yet. Pass channel to keep one channel's sites and pages; the totals then count that channel only.

list_projects

ChatGPT
Check in on every Frontmost AI project. Returns Visibility, Position, and Sentiment for each project's latest delivered scan, with counts and the change since the previous scan. metric_breakdown separates total, branded, and unbranded metrics and coverage. Use total and unbranded together for performance conclusions. Also returns running scans, open recommendations, the next scheduled scan, each project's run_config (AI apps, cadence, weekly brief, active questions, and the most credits a month uses), the questions in use, the projects the plan allows, and the workspace's credits: what is left, when they reset, and the projected use this month. Projects with the largest change come first. Use this tool first for any question about results. Copy a KPI's answers_filter into get_answers to read the answers behind it.

load_skills

ChatGPT
Load the full text of 1 to 5 skills by name: Frontmost AI's step-by-step guidance for a task, written for this connector's tools. Follow the text for the rest of the task. A prompt copied from the app names the skill to load. Returns each skill's name, title, version, and body as Markdown, and missing for names that do not exist.

rate_brief

ChatGPT
Answer "Was this brief useful?" for a written brief: Useful or Not useful, with an optional note on what would make the briefs more useful. A new rating replaces the earlier one. The feedback is used to improve future briefs.

search_industries

ChatGPT
Find six-digit NAICS 2022 industries by plain words, like dentist or project management software, or by code. Use a code with update_project.

search_skills

ChatGPT
Find Frontmost AI's guidance for a task, such as listing outreach, Reddit replies, or a page that answers a question. Every word of query must appear in a skill's name, title, or description, or in the title of an app prompt that uses it. Omit query to list every skill. Returns up to 10 skills with name, title, description, version, and used_by, without their text. Then call load_skills with the names.

share_scan

ChatGPT
Email a scan's results with the PDF report attached to up to 10 addresses, as the app's Share menu does. Replies go to the person who shares it. Omit scan_id for the latest delivered scan.

update_project

ChatGPT
Change a project's name, run_config (AI apps, scan cadence, weekly brief), weekly email, or industry. Each answer uses credits at its AI app's rate, and each weekly brief uses 25 credits, so more apps or a faster cadence use more of the monthly credits. The result's run_config gives the new cost. For industry_code, copy a six-digit code from search_industries. The result gives the plan's questions and projects in use.

update_recommendations

ChatGPT
Mark recommendations done or dismissed, or open them again, in one batch. Copy recommendation IDs from get_scan.

Capabilities

Writes

Example Prompts

Click any prompt to copy it.

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/frontmost-ai"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/frontmost-ai)