Nephia

Nephia is a brand monitoring service, and this is its remote MCP server.

От сообществаРаботаетБез входаГлобальныйБесплатноМожет изменять данные

Что умеет

  • List Mentions: Read everything your Queries caught, newest first, across every Source and every Query on the account — with the sentiment, intent, bucket and agent readings you have switched on. This
  • List Queries: List every Query on the account with what each caught in the last 24 hours (eventCount) and how much of it nothing has sorted yet (unclassifiedCount). Start here: a Query id is what get_
  • Get Query: Get one Query in full: what each of its searches actually polls (global criteria and per-source overrides already merged), its interval, whether AI reading and sentiment are on, its mute ru

Какие данные видит

Нужен ли аккаунт

Не нужен: сервер работает без входа

Nephia is a brand monitoring service, and this is its remote MCP server. Claude, Cursor, ChatGPT or any MCP client can read the mentions your brand gets on 14 sources: X, Reddit (posts and comments), YouTube, TikTok, Bluesky, Hacker News, Mastodon, Lemmy, GitHub, Product Hunt, Stack Overflow, any RSS feed, Vinted, and AI answers from ChatGPT, Gemini and Perplexity.

Every mention arrives already read, with its sentiment and intent, so an agent can answer plain questions: which complaints came in since Friday, what Reddit said about us this week. The source is an argument, not a tool, so one call reads every source you watch.

Sign-in is OAuth in the browser: no API key to copy. The consent screen has three permissions: read your mentions and Queries, change what is running (pause, resume, retire), and spend credits (semantic search and AI passes), which arrives unticked. Every tool description states its cost, so a model can budget before it spends.

The server is on every plan, Free included, and reading your own mentions through it costs nothing.

Как подключить

Как подключить Nephia к Claude

Адрес сервера, который нужно вставить в настройках Claude:

https://nephia.run.tools

  1. Откройте Claude (сайт claude.ai или приложение для компьютера).
  2. Перейдите в Настройки → Коннекторы (в английском интерфейсе: Settings → Connectors).
  3. Нажмите Добавить пользовательский коннектор (Add custom connector).
  4. Вставьте адрес сервера, скопированный ниже, в поле URL удалённого MCP-сервера и нажмите Добавить.
  5. В чате нажмите + → Коннекторы и включите новый коннектор.

Пользовательские коннекторы доступны на тарифах Free, Pro, Max, Team и Enterprise. На Free можно добавить только один. В Team и Enterprise коннектор сначала добавляет владелец организации в Настройки организации → Коннекторы.

Авторизация

После нажатия Добавить откроется окно входа в сервис. Войдите под своим аккаунтом и подтвердите доступ. Пароль Claude не видит.

Как подключить Nephia к ChatGPT

Адрес сервера, который нужно вставить в настройках ChatGPT:

https://nephia.run.tools

  1. Откройте ChatGPT в браузере (chatgpt.com). Нужен тариф Plus, Pro, Business, Enterprise или Edu.
  2. Один раз включите режим разработчика: Настройки → Приложения → Дополнительные настройки → Режим разработчика (Settings → Apps → Advanced settings → Developer mode).
  3. Откройте Настройки → Коннекторы и нажмите Создать (Create).
  4. Заполните форму: Название (любое), Описание (одна фраза, что делает сервис), URL MCP-сервера (скопируйте ниже).
  5. В поле Аутентификация выберите OAuth, если сервис требует вход, иначе Нет. Нажмите Создать.
  6. В новом чате откройте + → Приложения/Коннекторы и включите коннектор.

Название раздела у OpenAI менялось (Connectors → Apps/Plugins). Если пункт называется иначе, ищите «режим разработчика» в настройках. В рабочих пространствах Business/Enterprise администратор должен разрешить пользовательские коннекторы.

Авторизация

При первом использовании ChatGPT откроет окно входа в сервис. Войдите и подтвердите доступ.

Как подключить Nephia к Cursor

Адрес сервера, который нужно вставить в настройках Cursor:

https://nephia.run.tools

Быстрый способ: нажмите кнопку Открыть в Cursor ниже, Cursor сам предложит добавить сервер.

Вручную:

  1. В Cursor откройте Settings → Cursor Settings → MCP и нажмите Add new global MCP server.
  2. Вставьте JSON, скопированный ниже, в открывшийся файл ~/.cursor/mcp.json (для одного проекта: .cursor/mcp.json в корне проекта).
  3. Сохраните файл. Сервер появится в списке MCP; при необходимости пройдите авторизацию по кнопке рядом с ним.

Авторизация

Если сервис требует вход, рядом с сервером в списке MCP появится кнопка авторизации.

Список инструментов сервера (15)

Технические названия из tools/list. Нужны только разработчикам.

list_mentionsRead everything your Queries caught, newest first, across every Source and every Query on the account — with the sentiment, intent, bucket and agent readings you have switched on. This is the tool for questions like "what did Reddit say about us this week?", "any complaints since Friday?" or "show me negative mentions I have not answered". Free in the default text mode — these are your own rows. mode="semantic" asks the embedding index and charges 1 credit per question, then answers the same question free for 10 minutes, paging included. Do not set it to filter by keyword: that is what q= in text mode already does, for nothing. Scoped to one Query instead? Use get_query_results, which also carries each mention's bucket. Cursor-paged: pass nextCursor back unchanged.
list_queriesList every Query on the account with what each caught in the last 24 hours (eventCount) and how much of it nothing has sorted yet (unclassifiedCount). Start here: a Query id is what get_query_results, list_query_buckets and query_manage take. Free — does not charge credits.
get_queryGet one Query in full: what each of its searches actually polls (global criteria and per-source overrides already merged), its interval, whether AI reading and sentiment are on, its mute rules, its rules and the channels attached to it. Use it to answer "why did/didn't this get caught?" before assuming a Source is broken. Free — does not charge credits.
get_query_resultsThe same mentions as list_mentions, scoped to one Query and carrying the bucket each was sorted into. Use it for "what has my <brand> monitor caught?" and, with bucket=, for "show me the pricing complaints". Free in the default text mode — these are your own rows. mode="semantic" asks the embedding index and charges 1 credit per question, then answers the same question free for 10 minutes, paging included. Do not set it to filter by keyword: that is what q= in text mode already does, for nothing. Cursor-paged: pass nextCursor back unchanged.
list_query_bucketsThe buckets this Query sorts its mentions into — id, label, the plain-language rule behind it, and a count each — plus how many are unsorted and how many are still pending. The ids are what get_query_results takes as bucket=, and the labels are what the ids in its output mean. Free — does not charge credits. Listing buckets never triggers a sorting pass, so it cannot bill.
query_manageManage a Query's lifecycle across every Source at once: pause stops all its polling and all its spending, resume restarts it, delete retires it and frees the plan slot. Free — does not charge credits. Pausing is also how you stop the polling that does. On resume, a Source an operator stopped stays stopped and comes back named in blockedSources rather than failing the call — read that field before reporting success. delete takes an optional purge: without it the mentions the Query caught stay readable, with purge=true they are deleted too and cannot be recovered. Ask before pausing or deleting, and ask again before purging: this is the customer's monitoring and the mentions are the record of what it found.
watch_createCreate a Nephia watch that polls a source on an interval. webhookUrl is optional: with one, new matches are POSTed to a publicly reachable endpoint the user controls — ask for it, never invent one; without it, the watch records matches and you read them with watch_feed. Billing is metered per tick at the source's search credit rate: the watch keeps consuming credits until paused or deleted. source=vinted needs criteria; source=x needs query XOR terms; source=reddit needs query/terms and/or subreddit; source=youtube needs query/terms XOR channel; source=bluesky needs query and/or author; source=hackernews needs query; source=mastodon needs hashtag XOR account and takes no query — Mastodon has no global keyword search; source=lemmy needs query and/or community; source=github needs query XOR repo — query is GitHub's own search syntax, passed through, and repo polls that project's releases; its interval floor is 300s because the search budget is per token, not per watch; source=producthunt takes an optional topic and/or query and its query is matched over the newest page rather than searched upstream — the API has no text search on launches; source=stackoverflow needs query and takes optional tags, and its results include answers as well as questions; source=tiktok needs query XOR user and takes an optional region — a query matches video captions only (TikTok indexes speech and on-screen text, the API returns neither), a user polls that creator's uploads, and its interval floor is 3600s because every tick is bought upstream whether or not anything is new; source=rss needs feedUrl XOR query/terms; source=ai-answers needs prompt and takes no query/terms — it is asked, not searched, and its interval floor is 6h because every run is a paid prompt. terms carries several search texts on one watch for one tick's cost, on x/reddit/youtube/rss only. Pass ai to have every event read by an agent step (1 credit per 25 items). Pass backfill to seed the watch with one immediate search (one extra tick) so it is not empty on day one.
watch_listList the account's watches for a source, or get one watch by id. Free — does not charge credits.
watch_manageManage a watch's lifecycle: pause stops polling (and credit consumption), resume restarts it, delete deactivates the watch permanently. Free — does not charge credits. Pausing is also how you stop the polling that does.
watch_feedPoll a watch without a webhook receiver: kind=events replays matched events since an ISO timestamp (newest first); kind=activity shows poll heartbeats and webhook delivery outcomes (default window: last 7 days); kind=runs is source=ai-answers only and returns the stored answer runs with their citations and brand mentions. Free — does not charge credits.
analyses_runRead a set of items with one AI pass. Charges credits: 1 credit per 50 items for group, classify and summarise, and 1 credit per 25 for agent — the agent answers your own fields per item, which is heavier than a single verdict. The response's meta.credits_used is what was actually charged; a batch the provider could not answer is not billed. Estimate first with analyses_estimate if the cost matters.
analyses_estimateCosts 0 credits. Returns how many credits analyses_run would charge for that many items, per kind, computed by the same functions that charge it.
vinted_price_historyGet every price and status Nephia observed for one Vinted listing, oldest first. Costs 3 credits per request. Points are movements, not poll ticks: one is recorded when the price or status changed, or once every 24 hours while neither did. Coverage grows with usage: a series exists only for items that appeared in a search Nephia observed. This is not a complete index of Vinted.
vinted_market_statsGet p25 / median / p75 and a listing count per day for a Vinted market, newest first, filterable by brand, catalog, size, condition and currency. Costs 5 credits per request. Defaults to the last 30 days; a range may not exceed 365 days. soldRate is always null — sold detection is not yet available. Coverage grows with usage: a series exists only for items that appeared in a search Nephia observed. This is not a complete index of Vinted.
account_creditsGet the remaining credit balance for the Nephia account. Free — does not charge credits. Use this before a charged call, or after an insufficient-credits error.