MCP-сервер ScaledPapers

• Имя в реестреcom.scaledpapers/scaledpapersподтверждён в реестре MCP

Имя в реестре лежит в пространстве scaledpapers.com. Реестр пускает в такое пространство только того, кто доказал владение доменом через DNS или HTTP.

• Состояниеотвечаетпроверено 2026-10-30
• Инструменты34
• Сервер называет себяScaledPapers
• Версия протокола2025-06-18

2026-10-30 сервер ответил на initialize и tools/list без входа. Первый ответ за 37 мс.

• Делают ли инструменты работу продукта, ещё не проверено.
• Проверок пока 3 из 7, нужных, чтобы судить о надёжности.

Вывод сделан правилами по замерам выше. Вероятный первый вызов выбрала модель по списку инструментов: прочитайте описание, прежде чем на него полагаться.

• Оценка INITE MCPчерез неделюдля оценки нужно 7 ежедневных проверок, пока 3
• Отвечалидёт замерпроверок пока 3, долю покажем после 7
• Медиана первого ответа36 мс
• Место в контексте агента≈ 6 180 токеновописания всех инструментов (34), они занимают контекст агента на каждом ходу
• Описания инструментовB · 86 из 100
• Типизированные ответы6 из 34 инструментовобъявлен outputSchema: ответ приходит с типами и читается сразу

• Параметры без описания: в 28 из 34

Программа проверяет сервер раз в день: initialize и tools/list, без модели. Токены - это длина описаний из расчёта примерно четыре символа на токен. Оценка - среднее шести проверок имени, описания, схемы и аннотаций каждого инструмента. Она говорит, насколько понятно описаны инструменты, а не о том, работают ли они. Оценка складывается только из замеренного: отвечает ли сервер (25%), делает ли работу продукта (25%), описания (20%), контекст (10%), использование (10%) и насколько просто подключиться (10%). Незамеренная часть не учитывается, а скрытая инструкция ограничивает оценку 40 баллами. Состояние MCP-серверов вендоров: методика и цифры →

• usage

  Plan caps for this workspace: plan, period, probe_count, monthly_probe_quota, endpoints_used, max_endpoints, max_regions, min_interval_seconds, sample_retention_days, max_status_pages. Call this before creating anything. Stop when a cap…

• list_regions

  Enabled probe regions: id, name, and region.

• list_checks

  Checks in this workspace. Use this before create_check so you do not add a second check for a URL that is already watched.

• get_check

  One check, including assertion, DNS, and TCP fields. Secrets such as request headers, request body, and HTTP auth are not returned.

• get_check_results

  Check results. view is status, history, or outages. history uses hours (default 24) and returns status, status_code, latency_ms, dns_ms, connect_ms, location_name, location_region, and observed_at. Sample metadata is omitted.

• create_check

  Create a check. kind is http, api, dns, or tcp. Defaults: GET, status 200, 10s timeout, email alerts on, interval_seconds from the plan minimum, and one region. DNS defaults to an A record over UDP. TCP requires port. Pass regions only…

• update_check

  Update a check. Pass only the fields to change. enabled false pauses and true resumes. Linking requires site_id, site_slug, and component_id together, plus optional incident_mode (off, auto, or review: review stores a draft a person…

• delete_check

  Delete a check. Run only when the user asked to delete this check in this conversation.

• run_check

  Queue a probe for one check. Returns job_id. Call get_check_results with view status for the result; it is not included here.

• list_status_pages

  Status pages in this workspace: slug, name, visibility, publicUrl, custom domain, and the worst open incident status.

• get_status_page

  One status page: components with ids, groups, incidents with public ids, Slack connection state, and alert-destination ids. has_password is returned; the password is not.

• create_status_page

  Create a public status page. visibility internal requires Pro and a password of at least 8 characters that the user typed; never invent a password or repeat it. custom_domain is a hostname such as status.acme.com and requires Pro. With…

• update_status_page

  Update a status page. Fields: name, url, visibility, password, custom_domain, show_auto_detected (label incidents a check opened itself as detected automatically). custom_domain null removes it. visibility internal needs a password of at…

• delete_status_page

  Delete a status page and its components. Run only when the user asked to delete this page in this conversation.

• verify_custom_domain

  Check the CNAME and TXT records for a status page custom domain.

• create_component

  Add a component to an existing status page. Returns its id so update_check can link a check to it.

• update_component

  Rename a component or change its URL.

• delete_component

  Delete a component. Run only when the user asked to delete this component in this conversation.

• manage_component_group

  Create, update, or delete a component group. action is create, update, or delete. create and update need name. update and delete need group_id. Delete only when the user asked.

• slack_connection

  Slack connection for a status page. action start returns a URL for the user to open; do not complete OAuth. action disconnect removes the connection. Disconnect only when the user asked. Connection state is on get_status_page.

• page_alert_destinations

  Replace the alert destinations attached to a status page with these existing destination ids. Use list_alert_destinations for the ids. An empty list detaches all of them.

• list_incident_proposals

  Drafts waiting for review from checks in incident_mode review, pending first. Approving one publishes it as a status incident.

• approve_incident_proposal

  Publish a pending review draft as a status incident. This emails subscribers and posts to Slack. Run only when the user asked to publish it in this conversation. Optional title, body, and impact (minor, major, or critical) replace the…

• dismiss_incident_proposal

  Dismiss a pending review draft. Nothing is published.

• create_incident

  Open an incident. This emails subscribers and posts to Slack the same way the console does. Run only when the user asked to publish this incident in this conversation. status is investigating, identified, monitoring, or resolved. impact…

• get_incident

  One incident, including updates newest first.

• update_incident

  Edit an incident. This can email subscribers and post to Slack. Run only when the user asked. With update_id, edit that posted update instead of the incident. Never invent a status the user did not ask for.

• post_incident_update

  Post an incident update. This emails subscribers and posts to Slack. Run only when the user asked to publish this update in this conversation. status resolved closes the incident.

• list_maintenance_windows

  Maintenance windows: id, name, starts_at, duration_seconds, timezone, recurrence, weekdays, ends_at, notify_subscribers, and items.

• create_maintenance_window

  Schedule a maintenance window. duration_seconds is 60 through 7 days. Pass both the check (endpoint_ids) and its public component (status_components) so probes pause and the page shows Under maintenance. notify_subscribers defaults to…

• update_maintenance_window

  Update a maintenance window. notify_subscribers defaults to false unless you set it. This can email subscribers. Run only when the user asked to change this window.

• cancel_maintenance_window

  Cancel a maintenance window. notify_subscribers defaults to false. This can email subscribers. Run only when the user asked to cancel this window in this conversation.

• list_alert_destinations

  Existing alert destinations: id, kind, name, and hint. Secrets are not returned. Creating a destination stays in the console.

• test_alert_destination

  Send a real test alert to an existing destination. Run only when the user asked to send the test.