MCP-сервер Webhook Studio

• Имя в реестреcom.webhook-studio/webhooksподтверждён в реестре MCP

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

• Состояниеотвечаетпроверено 2026-10-25
• Инструменты17
• Сервер называет себяwebhook-studio
• Версия сервера0.1.1
• Версия протокола2025-06-18

2026-10-25 сервер ответил на initialize и tools/list без входа. Первый ответ за 86 мс.

• Инструменты делают все проверенные сценарии: capture and inspect incoming provider webhook requests; generate typescript types from real webhook payloads; replay captured events to localhost with original signatures.
• Проверок пока 1 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 1
• Что агент может с ним сделать3 из 3 сценариев
• Отвечалидёт замерпроверок пока 1, долю покажем после 7
• Медиана первого ответа232 мс
• Место в контексте агента≈ 3 161 токеновописания всех инструментов (17), агент держит их в контексте на каждом ходу
• Описания инструментовB · 83 из 100
• Типизированные ответы0 из 17 инструментовобъявлен outputSchema: ответ приходит с типами и читается сразу

• Не сказано, только читает инструмент или может удалять: 17 из 17

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 (30%), делает ли работу продукта (25%), описания (20%), контекст (10%) и использование (15%). Незамеренная часть не учитывается, а скрытая инструкция ограничивает оценку 40 баллами. Состояние MCP-серверов вендоров: методика и цифры →

• create_bucket

  Provision a new webhook endpoint (a "bucket") that captures incoming HTTP requests. Returns its id, public ingest url to point a provider at, inbox_url (the live human view to hand back), and signing secret. Pass external_ref — your own…

• list_buckets

  List your webhook endpoints. Pass external_ref to fetch the specific bucket you created earlier under that name — this is how you re-anchor after losing the conversation, since you never need to have stored the bucket id.

• get_bucket

  Fetch one webhook endpoint by id, including its url and inbox_url.

• delete_bucket

  Permanently delete a webhook endpoint and its captured events.

• latest_event

  Get the single most recent event captured by a bucket, optionally filtered by provider and event type. Returns one event with its headers, body and delivery results — the one-shot answer to "what did Stripe last send me". 404 if nothing…

• list_events

  List events captured by a bucket, newest first, with composable filters: since/until (ISO 8601 or YYYY-MM-DD), provider, type, method, signature_valid, and q (case-insensitive substring over body, content type, method, type and labels)…

• get_event

  Fetch one captured event by id with its full headers, body, signature verification result (and failure reason), and every outbound delivery attempt. This is how you find out whether your forward actually reached its target and what came…

• wait_for_event

  Block until the next matching event arrives on a bucket, or until the timeout. Use this to turn "go click Send test webhook in Stripe" into a synchronous step: call it, tell the human to trigger the event, and it returns as soon as one…

• replay_event

  Re-send a captured event to any URL — the test runner for a handler you just wrote. preserve_headers defaults to true so the original signature header (e.g. Stripe-Signature) arrives intact. If you changed the body or the receiver…

• forward_bucket

  Set up standing forwarding from a bucket to a URL. Pass forward_all:true to forward every event, or condition_tree to forward only matching events (same AND/OR condition grammar the routing engine uses). Creates the destination and…

• list_deliveries

  List outbound delivery attempts for a bucket, newest first, with the full request and response of each. Pass success:false to see only failures — the direct answer to "which of my forwards are broken and why".

• configure_verification

  Turn on HMAC signature verification for a bucket using the provider's own signing secret (e.g. Stripe's whsec_). This is how you finish a "verified webhook" integration end to end. Set secret to the provider secret and hmac_enabled:true…

• list_schemas

  List the payload schemas Webhook Studio has learned from this account's real traffic, optionally filtered by provider and event_type. These outlive the payloads they were learned from, so this answers "what shapes do I know about" even…

• get_schema

  Fetch one learned schema with every field, its type, whether it is optional, and how often it actually appears — the real shape of the payload, derived from what this account received rather than guessed from training data.

• diff_schema

  Compare two versions of a learned schema and get exactly which fields were added, removed or changed type, each flagged as breaking or not. This is the "why did my handler start failing" answer, produced without reading a single payload…

• generate_types

  Generate types, a runtime validator, or a full handler from a learned schema — the payoff of the knowledge layer. Everything is derived from the payloads this account ACTUALLY received, with genuinely-optional fields marked optional…

• find_correlation_keys

  List the correlation-key candidates for a learned schema — the fields that look like identifiers (e.g. data.object.id), ranked by how consistently they appear. These are the fields to group events by when reconstructing a lifecycle (a…