tracker-mcp
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@tracker-mcp найди мои открытые задачи в Трекере"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
traker-mcp — MCP-сервер для Яндекс Трекера
Проще говоря: этот сервер учит вашего AI-ассистента работать с Яндекс Трекером. Вы говорите ассистенту «найди мои задачи», «создай задачу в очереди TREK», «закрой TASK-123 с резолюцией “Решён”» — а он делает это через API Трекера сам, без копирования ссылок руками.
Работает с любым MCP-клиентом: VS Code (Copilot), Cursor, Claude Desktop, Claude Code, opencode, Windsurf, Zed, JetBrains AI Assistant, Grok CLI.
📘 Подробное руководство для новичков (получение токена, пошаговая настройка каждого клиента, решение проблем) — в docs/SETUP.md.
Документация
Сайт документации (docs7) и LLM-индекс на Context7: context7.com/sdamarketing/tracker_mcp.
Раздел | Содержание |
Обзор, быстрый старт, установка, клиенты | |
Гид по возможностям | Задачи · Комментарии/чеклисты/worklog/файлы · Очереди · Доски и спринты · Проекты/цели/портфели · Автоматизация · Админ · Архитектура |
Справочник |
Отдельно: CHANGELOG и skills.sh-скилл.
Related MCP server: YaTracker Connector
Ресурсы и промпты (MCP)
Помимо инструментов сервер отдаёт:
Resources (
resources/list,resources/read) — tracker:// URI для чтения контекста без вызова инструментов:tracker://issue/{KEY}— карточка задачи с описанием и последними 20 комментариями (markdown);tracker://queue/{KEY}— очередь + обязательные поля создания задач;tracker://board/{ID}— доска с колонками и статусами;tracker://catalog/statuses|priorities|types|resolutions|linktypes— справочники (делят TTL-кэш с инструментами);tracker://myself— ваш профиль.
Prompts (
prompts/list,prompts/get) — готовые сценарии с аргументами:standup,weekly-report,triage,sprint-review,close-issue,create-issue. В клиентах с поддержкой промптов вызываются как slash-команды.
Что умеет
187 инструментов — полное покрытие API Трекера v3:
Категория | Что можно делать |
Задачи | искать (фильтр, язык запросов, query 2.0, scroll >10k, подсказки), создавать, читать, редактировать, переносить, менять статус, связывать, история |
Комментарии | добавлять, читать (список и по одному), редактировать, удалять, реакции |
Чеклисты | задачи + проекты/портфели: добавлять, отмечать, двигать, удалять |
Учёт времени | записи по задаче и поиск по авторам/датам |
Вложения | список, метаданные, скачивание (картинки/текст/база), миниатюры, загрузка, временные файлы |
Массово | редактирование, перенос и закрытие задач; bulk-обновление проектов/целей |
Отчёты и фильтры | сохранённые отчёты для виджетов, CRUD сохранённых фильтров |
Очереди | CRUD (вкл. удаление/восстановление), права доступа, версии, компоненты, локальные поля, макросы, теги |
Автоматизация | триггеры и автодействия (CRUD + логи), воркфлоу (статусы и переходы) |
Доски | CRUD, пагинация, колонки CRUD, спринты (создать/запустить/архив) |
Дашборды | создание дашбордов и виджетов («Время цикла») |
Проекты, цели, портфели | CRUD, комментарии, чеклисты, вложения, связи, события, права, ключевые результаты и метрики (через fields) |
Справочники | чтение И запись: типы, статусы, приоритеты, резолюции; глобальные поля и категории |
Внешние приложения | связи задач с внешними системами (Jira и др.) |
Отсутствия | отпуска/болезни/командировки (admin) |
Миграция | импорт задач/комментариев/связей/трудозатрат/вложений «задним числом» (admin) |
Безопасность для агентов
Каждый инструмент несёт MCP-аннотации: чтения помечены
readOnlyHint, всеdelete_*—destructiveHint— клиенты показывают их как подтверждаемые/Безопасные операции.Все
delete_*иbulk_*требуютconfirm: true(агент сначала показывает, что удалит).TRACKER_READ_ONLY=1— сервер отдаёт только 80 инструментов чтения, запись скрыта полностью. Для «просто посмотреть» интеграций.
Скилл для AI-агентов (skills.sh)
К серверу прилагается скилл yandex-tracker — процедурные знания для агента:
язык запросов Трекера, корректные значения полей (приоритеты, статусы, резолюции),
рецепты («найди мои задачи», «закрой с резолюцией», bulk-правки) и грабли API
(dueDate vs releaseDate, обязательная резолюция при закрытии, IAM-токены ≤12ч).
npx skills add sdamarketing/tracker_mcpУстанавливает скилл во все обнаруженные агенты (Claude Code, Cursor, opencode,
Codex и ещё 75+). Мастер npm run setup тоже предлагает установить скилл
вместе с настройкой MCP-клиента. Репозиторий скилла: skills.sh/sdamarketing/tracker_mcp.
Установка в одну команду (macOS / Linux / WSL)
curl -fsSL https://raw.githubusercontent.com/sdamarketing/tracker_mcp/main/install.sh | bashСкрипт сам: проверит систему → поставит Node.js через nvm, если его нет
(спросит согласия, sudo не нужен) → скачает сервер в ~/.traker-mcp → соберёт
и проверит его → запустит мастер настройки: ключи Трекера (ввод скрыт),
проверка ключей в API, выбор AI-клиента (VS Code, Cursor, Claude Desktop,
Claude Code, opencode, Windsurf, Zed, JetBrains, Grok CLI). Если команда grok
уже есть, мастер прописывает сервер в ~/.grok/config.toml сразу, до меню.
Повторный запуск той же командой = обновление сервера (конфиги клиентов не трогаются).
🔎 Хотите сначала прочитать скрипт? Уберите
| bashи посмотрите вывод:curl -fsSL https://raw.githubusercontent.com/sdamarketing/tracker_mcp/main/install.sh🪟 Windows: установите через WSL командой выше или вручную — см. ниже.
Установка из npm (без клонирования)
npm install -g tracker-mcp
tracker-mcp setup # интерактивный мастер: ключи Трекера + настройка AI-клиентаПосле этого сервер доступен командой tracker-mcp (в конфиг AI-клиента её и
прописывайте). Другие команды: tracker-mcp update, tracker-mcp links,
tracker-mcp --help.
Official MCP Registry
Сервер опубликован в официальном реестре MCP как
io.github.sdamarketing/tracker-mcp (npm + Docker-пакеты, env-переменные описаны
в server.json). Клиенты с поддержкой реестра найдут его по имени.
Docker (без установки чего-либо, кроме Docker)
Образ публикуется в GitHub Container Registry при каждом релизе (v*-тег):
# stdio (MCP-клиент запускает контейнер сам)
docker run -i --rm -e TRACKER_TOKEN -e TRACKER_ORG_ID ghcr.io/sdamarketing/tracker-mcp
# HTTP-режим
docker run --rm -p 3407:3407 -e TRACKER_TOKEN -e TRACKER_ORG_ID \
-e MCP_AUTH_TOKEN=вашключ ghcr.io/sdamarketing/tracker-mcp serve --host 0.0.0.0В конфиге агента вместо command: node используется
command: docker, args: ["run","-i","--rm","-e","TRACKER_TOKEN","-e","TRACKER_ORG_ID","ghcr.io/sdamarketing/tracker-mcp"].
Ручная установка (Windows или без curl)
Шаг 1. Убедитесь, что есть Node.js 20+ (проверка: node -v).
Шаг 2. Соберите сервер:
git clone <адрес-этого-репозитория>
cd traker_mcp
npm install
npm run build
npm run smoke # проверка: должно быть "OK: ... 187 tools listed"Шаг 3. Подготовьте две вещи из Трекера:
токен — как получить: docs/SETUP.md, раздел 3
ID организации — как найти: docs/SETUP.md, раздел 4
Шаг 4. Запустите мастер установки:
npm run setupМастер спросит ключи (токен — скрытым вводом), проверит их в API Трекера, покажет интерактивное меню клиентов (↑↓ + Enter) и настроит выбранные (можно несколько подряд) — через их CLI или дописав конфиг с бэкапом, — а в конце выведет отчёт.
Альтернатива: npm run links — откроет страницу в браузере с кнопками
установки в один клик (VS Code, Cursor, Claude Desktop) и конфигами всех
клиентов для ручного копирования.
Подключение к вашему агенту
Ваш агент | Быстрый способ | Конфиг-файл |
VS Code (Copilot) |
|
|
Cursor |
|
|
Claude Desktop |
|
|
Claude Code (терминал) |
|
|
opencode |
|
|
Windsurf |
|
|
Zed |
|
|
JetBrains IDE |
|
|
Grok CLI |
|
|
Пошаговые инструкции с точными путями для macOS/Windows/Linux: docs/SETUP.md.
Проверка
После установки перезапустите агента и спросите:
«Вызови инструмент get_current_user» — вернётся ваш профиль из Трекера. «Найди мои задачи в Трекере» — сработает find_issues.
Частые проблемы
Симптом | Причина и решение |
| токен не передан — перегенерируйте ссылки ( |
| неверный токен или токен истёк (IAM живёт ≤12 часов) |
| неверный |
Сервер не запускается в Claude Desktop (macOS) | используйте абсолютный путь к node — уже так в сгенерированном конфиге |
У агента нет инструментов | сервер отключён — включите в списке MCP-серверов клиента и перезапустите |
Больше решений — docs/SETUP.md, раздел 8.
Настройка сервера
Переменная | Обязательна | Описание |
| да | OAuth-токен ( |
| да | Идентификатор организации |
| нет |
|
| нет | По умолчанию |
| нет |
|
| нет |
|
| нет | TTL кэша справочников (по умолчанию 10 минут) |
HTTP-режим (serve)
По умолчанию сервер работает по stdio (для локальных агентов). Для сетевого доступа (команда, удалённый клиент, проксирование):
tracker-mcp serve --port 3407 --host 127.0.0.1 # значения по умолчанию
MCP_AUTH_TOKEN=s3cret tracker-mcp serve # ключ к эндпоинту (Bearer)
curl localhost:3407/health # {"ok":true,"tools":187}Эндпоинт MCP:
POST/GET/DELETE /mcp(Streamable HTTP, stateless)Без
MCP_AUTH_TOKENпривязка к не-loopback-адресу отклоняется — токен Трекера живёт в env, голый порт в сети недопустимКонфиг клиента для HTTP:
"url": "http://127.0.0.1:3407/mcp"+"headers": {"Authorization": "Bearer s3cret"}
Безопасность
Токен даёт агенту те же права, что у вас в Трекере. Выдавайте минимальные права (для чтения хватит
tracker:read).Храните токен в
.env(уже в.gitignore) или в конфиге клиента, не в коде.install-links.htmlсодержит токен — скрипт добавляет его в.gitignoreсам.
Разработка
npm run typecheck # проверка типов
npm run build # сборка в dist/
npm run smoke # запуск и список инструментов без обращения к API
npm run setup # интерактивный мастер установки для AI-клиентов
npm run links # страница с кнопками установки (deeplinks)Архитектура и полный список инструментов — в docs/SETUP.md и в коде src/tools/.
Лицензия и поддержка
Проблемы и идеи — в issues репозитория. API Трекера меняется — следите за официальной документацией.
Available Tools
187 toolsadd_checklist_itemC
Add a checklist item to an issue.
Inputs: text (required), issueId (required), checked, assignee, +1 more.
add_commentB
Add a comment to an issue. Supports Yandex Flavored Markdown.
Inputs: text (required), issueId (required).
add_comment_reactionC
Add a reaction to a comment.
Inputs: issueId (required), reaction (required), commentId (required).
add_entity_attachmentA
Attach a temporary file (from upload_temp_attachment) to a project, portfolio or goal.
Inputs: entityId (required), entityType (required), tempFileId (required), notify.
add_entity_checklist_itemsC
Create a checklist or append items to a project/portfolio checklist.
Inputs: items (required), entityId (required), entityType (required), notify.
add_entity_commentB
Add a comment to a project, portfolio or goal.
Inputs: text (required), entityId (required), entityType (required).
add_worklog_recordB
Add a time tracking record to an issue. Tracker counts time in 5-day work weeks and 8-hour work days: P5D is displayed as 1 week.
Inputs: start (required), issueId (required), duration (required), comment.
archive_sprintB
Archive a finished sprint (version = sprint version).
Inputs: version (required), sprintId (required).
bulk_move_issuesC
Bulk move issues to another queue. Components, versions and projects are cleared by default.
Inputs: queue (required), issues (required), notify, confirm.
bulk_transition_issuesA
Bulk move issues to a new status via a workflow transition. Get available transition ids via get_issue_transitions. Statuses like "closed" may require a resolution in values.
Inputs: issues (required), transition (required), notify, values, +1 more.
bulk_update_entitiesC
Bulk update projects, portfolios or goals: fields, comment and links at once.
Inputs: entityType (required), metaEntities (required), links, fields, +2 more.
bulk_update_issuesB
Bulk edit many issues at once (max 10000). Values use the issue edit format; array fields support add/set/replace commands, e.g. {"tags": {"add": ["x"]}}.
Inputs: issues (required), values (required), notify, confirm.
count_issuesBRead-only
Get the number of issues matching a filter or query.
Inputs: query, queue, filter.
create_boardB
Create a new task board (via the liveBoards API; the legacy POST /boards ignores body parameters). Columns map board columns to issue status keys.
Inputs: name (required), owner, columns, autoFilters, +3 more.
create_board_columnA
Create a column on a board. version = current board version (If-Match lock; get it from get_board).
Inputs: name (required), boardId (required), version (required), statuses (required).
create_componentC
Create a component in a queue.
Inputs: name (required), queue (required), lead, assignAuto, +1 more.
create_cycle_time_widgetA
Add a "Cycle time" chart widget to an existing dashboard. Issue source priority: filterId > query > filter.
Inputs: dashboardId (required), description (required), end, mode, +11 more.
create_dashboardC
Create a new dashboard.
Inputs: name (required), owner, layout.
create_entityC
Create a project, portfolio or goal.
Inputs: summary (required), entityType (required), links, fields.
create_entity_linkB
Link a project/portfolio/goal to other entities. Relationships: projects/portfolios "depends on", "is dependent by", "works towards" (→ goal); goals "parent entity", "child entity", "is supported by" (← project).
Inputs: links (required), entityId (required), entityType (required).
create_external_linkA
Link an issue to an object in an external application (e.g. Jira via a connector). Use relationship "RELATES" unless the app docs say otherwise.
Inputs: key (required), origin (required), issueId (required), relationship (required), +1 more.
create_filterC
Create a saved filter for issue lists.
Inputs: name (required), query, sorts, fields, +3 more.
create_gapsB
Create employee absences (vacation, illness, trip…). Admin rights required.
Inputs: gaps (required).
create_issueB
Create a new issue in Yandex Tracker. Requires queue key and summary. Text fields support Yandex Flavored Markdown.
Inputs: queue (required), tags, type, author, +18 more.
create_issue_fieldC
Create a global issue field (admin). type is a class name like "ru.yandex.startrek.core.fields.StringFieldType" (also Date, DateTime, Text, Float, Integer, User, Uri, Money, TimeTrackingDuration, StringFieldType with container=true).
Inputs: id (required), name (required), type (required), category (required), +7 more.
create_issue_field_categoryC
Create an issue field category (admin).
Inputs: name (required), order (required), description.
create_issue_linkC
Create a link between the current issue and another issue.
Inputs: issue (required), issueId (required), relationship (required).
create_issue_reportA
Create a saved issue-list report (used by dashboard "issue list" widgets). Pick exactly one issue source inside parameters.filter: query | filter | filterId.
Inputs: summary (required), query, sorts, fields, +3 more.
create_issue_typeB
Create a new issue type (admin).
Inputs: key (required), name (required).
create_priorityC
Create a priority (admin).
Inputs: key (required), name (required), order (required), description (required).
create_projectC
Create a project.
Inputs: name (required), lead, tags, queues, +3 more.
create_queueB
Create a new queue. Requires key, name, lead, default type/priority and an issueTypesConfig array binding issue types to workflow templates.
Inputs: key (required), lead (required), name (required), defaultType (required), +2 more.
create_queue_autoactionB
Create an autoaction: periodically updates issues matching filter/query. Actions are trigger-style objects (Transition, Update, CreateComment, Webhook, CalculateFormula).
Inputs: name (required), actions (required), queueId (required), query, +5 more.
create_queue_local_fieldC
Create a local (queue-specific) issue field.
Inputs: id (required), name (required), type (required), queueId (required), +7 more.
create_queue_macroB
Create a queue macro: comment body and/or issue field updates applied on run.
Inputs: name (required), queueId (required), body, issueUpdate.
create_queue_triggerB
Create a queue trigger. actions: Transition/Update/CreateComment/Webhook/CalculateFormula/… objects; conditions: field/event checks, single object {"type": "Or"|"And", "conditions": […]}.
Inputs: name (required), actions (required), queueId (required), active, +1 more.
create_queue_versionC
Create a version in a queue.
Inputs: name (required), queueId (required), dueDate, startDate, +1 more.
create_resolutionC
Create a resolution (admin).
Inputs: key (required), name (required).
create_sprintA
Create a sprint on a board (the board must have sprints enabled).
Inputs: name (required), boardId (required), endDate (required), startDate (required).
create_statusB
Create an issue status (admin).
Inputs: key (required), name (required), type (required).
create_workflowB
Create a workflow (admin): statuses/steps plus transitions. Omit "queue" to create a common workflow available to all queues. Attach issue types via issueTypeResolutions.
Inputs: name (required), steps (required), initialAction (required), id, +3 more.
delete_boardCDestructive
Delete a board by id.
Inputs: boardId (required), confirm.
delete_board_columnBDestructive
Delete a board column (version = board version).
Inputs: boardId (required), version (required), columnId (required), confirm.
delete_checklistBDestructive
Delete the whole checklist of an issue.
Inputs: issueId (required), confirm.
delete_checklist_itemBDestructive
Delete a checklist item from an issue.
Inputs: itemId (required), issueId (required), confirm.
delete_commentCDestructive
Delete a comment from an issue.
Inputs: issueId (required), commentId (required), confirm.
delete_componentCDestructive
Delete a component by id.
Inputs: componentId (required), confirm.
delete_entityBDestructive
Delete a project, portfolio or goal by id.
Inputs: entityId (required), entityType (required), confirm.
delete_entity_attachmentBDestructive
Delete a file from a project, portfolio or goal.
Inputs: entityId (required), entityType (required), attachmentId (required), confirm.
delete_entity_checklistBDestructive
Delete all checklist items of a project/portfolio.
Inputs: entityId (required), entityType (required), notify, confirm.
delete_entity_checklist_itemCDestructive
Delete one checklist item of a project/portfolio.
Inputs: itemId (required), entityId (required), entityType (required), notify, +1 more.
delete_entity_commentCDestructive
Delete a comment of a project, portfolio or goal.
Inputs: entityId (required), commentId (required), entityType (required), notify, +2 more.
delete_entity_linkBDestructive
Unlink an entity from another entity (by the linked entity id).
Inputs: right (required), entityId (required), entityType (required), confirm.
delete_external_linkCDestructive
Delete an external application link of an issue.
Inputs: linkId (required), issueId (required), confirm.
delete_filterCDestructive
Delete a saved filter (uses the v2 API path per official docs).
Inputs: filterId (required), confirm.
delete_gapsBDestructive
Delete absences by ids (admin).
Inputs: gapIds (required), confirm.
delete_issue_attachmentCDestructive
Delete an attached file from an issue.
Inputs: issueId (required), attachmentId (required), confirm.
delete_issue_linkADestructive
Delete a link between issues by link id (see get_issue_links).
Inputs: linkId (required), issueId (required), confirm.
delete_projectADestructive
Delete a project by id. Issues are not deleted.
Inputs: projectId (required), confirm.
delete_queueADestructive
Delete a queue (only when empty of issues). Only queue owner or Tracker admin can.
Inputs: queueId (required), confirm.
delete_queue_macroDDestructive
Delete a queue macro.
Inputs: macroId (required), queueId (required), confirm.
delete_queue_tagADestructive
Remove a tag from a queue (must not be used in any issue; admin only).
Inputs: tag (required), queueId (required), confirm.
delete_queue_versionCDestructive
Delete a queue version by id.
Inputs: versionId (required), confirm.
delete_sprintCDestructive
Delete a sprint by id.
Inputs: sprintId (required), confirm.
delete_workflowBDestructive
Delete a workflow by id (admin).
Inputs: workflowId (required), confirm.
delete_worklog_recordCDestructive
Delete a time tracking record (worklog).
Inputs: issueId (required), worklogId (required), confirm.
download_entity_attachmentARead-only
Download an attachment of a project/portfolio/goal. Content path for entities is /attachments/{id}/{filename} (not issue-scoped). Images come back as image content.
Inputs: attachmentId (required), filename.
download_issue_attachmentARead-only
Download attachment content: images come back as image content (visible in chat), text files as text, other binaries as base64. Omit filename to auto-resolve it by id.
Inputs: issueId (required), attachmentId (required), filename.
edit_checklist_itemA
Edit a single checklist item. The Tracker API requires the full item list in the request body, so this tool fetches the current checklist, merges your change into the target item, and sends the whole list back.
Inputs: itemId (required), issueId (required), text, checked, +2 more.
edit_commentC
Edit an existing comment text.
Inputs: text (required), issueId (required), commentId (required).
edit_issueA
Edit an existing issue (PATCH): only provided fields are changed. For array fields you may pass "add"/"remove"/"set"/"replace" commands via extraFields, e.g. extraFields={"tags": {"add": ["new-tag"]}}.
Inputs: issueId (required), tags, type, author, +18 more.
edit_worklog_recordC
Edit a time tracking record (worklog).
Inputs: issueId (required), worklogId (required), comment, duration.
execute_transitionA
Move an issue to a new status by executing a workflow transition. First get available transitions via get_issue_transitions. Some transitions (e.g. "close") require a resolution via extraFields, e.g. {"resolution": "fixed"}.
Inputs: issueId (required), transitionId (required), comment, extraFields.
find_gapsARead-only
Find absences of given users in a date range (admin). Pass the last user uid from the previous page to page through large organizations.
Inputs: users (required), to, from, page, +1 more.
find_issue_reportsARead-only
Search saved issue reports (filter by id/shortId/author).
Inputs: page, filter, orderBy, perPage, +1 more.
find_issuesARead-only
Find issues in Yandex Tracker. Pass exactly ONE search parameter: "queue" (queue key), "keys" (list of issue keys), "filter" (field->value object), "query" (Tracker query language string, e.g. "Queue: TREK Assignee: me()"), "query2" (query language 2.0 object) or "filterId" (saved filter id). Supports pagination via perPage/page.
Inputs: keys, page, order, query, +11 more.
find_worklog_recordsBRead-only
Search worklog records across all issues by author and/or creation date range.
Inputs: createdBy, createdTo, createdFrom.
get_attachment_thumbnailBRead-only
Download the preview thumbnail of an attached image (image content).
Inputs: issueId (required), attachmentId (required).
get_boardBRead-only
Get board parameters by id, including its columns.
Inputs: boardId (required).
get_board_columnCRead-only
Get parameters of one board column.
Inputs: boardId (required), columnId (required).
get_board_columnsCRead-only
List columns of a board.
Inputs: boardId (required).
get_boardsBRead-only
List all issue boards of the organization.
Inputs: perPage.
get_boards_paginateARead-only
List boards with relative pagination (for more than one page): pass the last board id as "id" to get the next page. Max 500 per page.
Inputs: id, perPage.
get_board_sprintsARead-only
List sprints of a board (agile boards only).
Inputs: boardId (required).
get_bulk_operation_infoARead-only
Get status and progress of a bulk change operation by its id.
Inputs: bulkOperationId (required).
get_checklistCRead-only
Get the checklist of an issue.
Inputs: issueId (required).
get_componentCRead-only
Get one component by id.
Inputs: componentId (required), fields.
get_component_group_accessCRead-only
Get group permissions for a component.
Inputs: groupId (required), componentId (required).
get_componentsARead-only
List all components of the organization.
Inputs: no named parameters.
get_component_user_accessCRead-only
Get user permissions for a component.
Inputs: userId (required), componentId (required).
get_current_userARead-only
Get info about the current user (the token owner). Good for connectivity checks.
Inputs: no named parameters.
get_entityARead-only
Get a project, portfolio or goal by id (use search_entities to find ids).
Inputs: entityId (required), entityType (required), fields.
get_entity_accessBRead-only
Get extended access settings of an entity, including inheritance sources.
Inputs: entityId (required), entityType (required).
get_entity_attachmentBRead-only
Get metadata of one file attached to a project, portfolio or goal.
Inputs: entityId (required), entityType (required), attachmentId (required).
get_entity_commentCRead-only
Get one comment of a project, portfolio or goal.
Inputs: entityId (required), commentId (required), entityType (required), expand.
get_entity_commentsCRead-only
List comments of a project, portfolio or goal.
Inputs: entityId (required), entityType (required), page, perPage.
get_entity_eventsBRead-only
Event history of a project, portfolio or goal (relative pagination: pass the last event id as "from" or focus with "selected").
Inputs: entityId (required), entityType (required), from, perPage, +3 more.
get_entity_linksCRead-only
List entity links of a project, portfolio or goal.
Inputs: entityId (required), entityType (required), fields.
get_entity_permissionsBRead-only
Get direct access permissions (READ/GRANT/WRITE) of an entity.
Inputs: entityId (required), entityType (required).
get_external_applicationsARead-only
List external applications available for issue external links.
Inputs: no named parameters.
get_external_linksARead-only
List external application links of an issue.
Inputs: issueId (required).
get_filterBRead-only
Get saved filter parameters by id.
Inputs: filterId (required).
get_issueCRead-only
Get issue parameters by key or id, e.g. "TREK-123".
Inputs: issueId (required), expand, fields.
get_issue_attachmentBRead-only
Get metadata of one attached file (name, size, mime type).
Inputs: issueId (required), attachmentId (required).
get_issue_changelogCRead-only
Get the change history (changelog) of an issue.
Inputs: issueId (required), page, perPage.
get_issue_commentBRead-only
Get one comment of an issue by id (numeric id or longId).
Inputs: issueId (required), commentId (required), expand.
get_issue_commentsCRead-only
List comments of an issue.
Inputs: issueId (required), page, perPage.
get_issue_fieldARead-only
Get parameters of a single global issue field by its key (e.g. "tags", "priority").
Inputs: fieldId (required).
get_issue_fieldsARead-only
List all global issue fields (standard and custom) with their types and options — useful to discover field ids for filters, create and edit operations.
Inputs: no named parameters.
get_issue_linksBRead-only
List links of an issue (related, dependent, subtask, duplicate, epic links). Pass perPage/page for the paginated variant.
Inputs: issueId (required), page, fields, perPage, +1 more.
get_issue_transitionsARead-only
List workflow transitions available for the issue (with target statuses). Use transition ids with execute_transition.
Inputs: issueId (required).
get_issue_typesARead-only
List all issue types (task, bug, epic, ...) available in the organization.
Inputs: no named parameters.
get_issue_worklogBRead-only
Get time tracking records (worklog) of an issue.
Inputs: issueId (required).
get_link_typesARead-only
List issue link types (relates, depends, subtask, duplicates, epic, clone) with their inward/outward names. Undocumented but live endpoint GET /v3/linktypes.
Inputs: no named parameters.
get_prioritiesBRead-only
List all issue priorities.
Inputs: no named parameters.
get_projectCRead-only
Get project parameters by id.
Inputs: projectId (required).
get_project_queuesBRead-only
List queues whose issues are included in a project.
Inputs: projectId (required), expand.
get_projectsBRead-only
List all projects of the organization.
Inputs: page, perPage.
get_queueCRead-only
Get queue parameters by key or id.
Inputs: queueId (required).
get_queue_autoactionCRead-only
Get parameters of one autoaction.
Inputs: queueId (required), autoactionId (required).
get_queue_autoaction_logsARead-only
Autoaction run logs. Without runId — list of launches; with runId — one launch with per-issue results (available for auto-update type autoactions).
Inputs: queueId (required), autoactionId (required), runId.
get_queue_autoactionsCRead-only
List autoactions configured in a queue.
Inputs: queueId (required).
get_queue_componentsBRead-only
List components of a queue.
Inputs: queueId (required).
get_queue_fieldsARead-only
Get fields available/required for issues of the queue (includes required flags and allowed values) — useful before creating an issue in this queue.
Inputs: queueId (required).
get_queue_group_accessCRead-only
Get group permissions in a queue.
Inputs: groupId (required), queueId (required).
get_queue_local_fieldCRead-only
Get one local field of a queue by its key.
Inputs: queueId (required), fieldKey (required).
get_queue_local_fieldsARead-only
List local (queue-specific) issue fields of a queue. Field ids look like "--" and can be used in create/edit issue requests.
Inputs: queueId (required).
get_queue_macroCRead-only
Get one macro of a queue.
Inputs: macroId (required), queueId (required).
get_queue_macrosesBRead-only
List macroes of a queue (saved bundles of field updates + comment text).
Inputs: queueId (required).
get_queuesCRead-only
List queues of the organization.
Inputs: page, perPage.
get_queue_tagsCRead-only
List tags used in a queue.
Inputs: queueId (required).
get_queue_triggerCRead-only
Get one trigger of a queue.
Inputs: queueId (required), triggerId (required).
get_queue_trigger_logsBRead-only
Webhook-action execution logs of a trigger (optionally for one issue, time range).
Inputs: queueId (required), triggerId (required), to, from, +2 more.
get_queue_triggersCRead-only
List triggers configured in a queue.
Inputs: queueId (required).
get_queue_user_accessCRead-only
Get user permissions in a queue.
Inputs: userId (required), queueId (required).
get_queue_versionCRead-only
Get one queue version by id.
Inputs: versionId (required), fields.
get_queue_versionsCRead-only
List versions of a queue.
Inputs: queueId (required).
get_queue_workflowsBRead-only
Get workflows of a queue: map workflow id → bound issue types.
Inputs: queueId (required).
get_resolutionsBRead-only
List all issue resolutions.
Inputs: no named parameters.
get_search_suggestARead-only
Autocomplete suggestions for issue summaries — use when the user references an issue by an imprecise name. Returns matching issues with keys.
Inputs: input (required), queue.
get_sprintCRead-only
Get sprint parameters by id.
Inputs: sprintId (required).
get_statusesBRead-only
List all issue statuses.
Inputs: no named parameters.
get_userCRead-only
Get user info by login or uid.
Inputs: login (required).
get_usersARead-only
List users of the organization (up to 10 000 — use get_users_relative beyond). Filter by email or group.
Inputs: id, email, group, expand, +1 more.
get_users_relativeARead-only
List users with relative pagination (works beyond the 10 000-user limit of get_users). Pass the last user uid as "id" to get the next page.
Inputs: id, expand, perPage.
get_workflowCRead-only
Get one workflow by id.
Inputs: workflowId (required).
get_workflowsCRead-only
List all workflows of the organization.
Inputs: no named parameters.
import_issueA
MIGRATION (admin): create an issue with a custom author and creation date. Use only when migrating from another tracker.
Inputs: queue (required), summary (required), createdAt (required), createdBy (required), +21 more.
import_issue_attachmentA
MIGRATION (admin): attach a local file with a custom author and date, to an issue or to one of its comments.
Inputs: issueId (required), filePath (required), createdAt (required), createdBy (required), +2 more.
import_issue_commentB
MIGRATION (admin): add a comment with a custom author and date.
Inputs: text (required), issueId (required), createdAt (required), createdBy (required), +2 more.
import_issue_linkA
MIGRATION (admin): create an issue link with a custom author and date. Supports all relationship types plus "clone" and "original".
Inputs: issue (required), issueId (required), createdAt (required), createdBy (required), +3 more.
import_worklog_recordB
MIGRATION (admin): add a worklog record with a custom author and date.
Inputs: start (required), issueId (required), duration (required), createdAt (required), +2 more.
list_entity_attachmentsCRead-only
List files attached to a project, portfolio or goal.
Inputs: entityId (required), entityType (required).
list_issue_attachmentsBRead-only
List files attached to an issue and its comments.
Inputs: issueId (required).
move_entity_checklist_itemC
Move a checklist item of a project/portfolio before another item.
Inputs: before (required), itemId (required), entityId (required), entityType (required), +1 more.
move_issueA
Move an issue to a different queue. Components, versions and projects are cleared by default (use moveAllFields to keep them). Local fields of the old queue are removed.
Inputs: queue (required), issueId (required), expand, notify, +2 more.
release_search_scrollA
Free scroll-search resources when you stopped paging through a scrolled find_issues result. Body: {"": ""} for every page you opened.
Inputs: scrollTokens (required).
restore_queueA
Restore a deleted queue (Tracker admin only).
Inputs: queueId (required).
search_entitiesARead-only
Search projects, portfolios or goals. Filter by name substring, field values or sort the result. Use entityType "goal" for OKR goals, "portfolio" for project portfolios.
Inputs: entityType (required), page, input, fields, +5 more.
set_queue_accessA
Change queue access rights: create/write/read/grant/deny with users, groups and roles. Arrays overwrite; {"add": […], "remove": […]} apply deltas. deny supports only users and groups.
Inputs: queueId (required), permissions (required).
start_sprintB
Start a sprint (moves to in_progress; version = sprint version).
Inputs: version (required), sprintId (required).
update_boardB
Update a board: name, backlog/sprints toggles, columns (with status mapping). Board owner cannot be changed via API.
Inputs: boardId (required), name, columns, backlogColumns, +3 more.
update_board_columnB
Rename a board column or change its statuses (version = board version).
Inputs: boardId (required), version (required), columnId (required), name, +1 more.
update_componentC
Edit a component: name, description, owner, default assignee. version = component version.
Inputs: version (required), componentId (required), lead, name, +2 more.
update_entityC
Update a project, portfolio or goal by id.
Inputs: fields (required), entityId (required), entityType (required), links, +1 more.
update_entity_accessA
Update extended access settings: enable/disable inheritance (permissionSources) and/or grant/revoke acl. Disable inheritance first — acl changes are rejected while inherited.
Inputs: entityId (required), entityType (required), acl, permissionSources.
update_entity_checklistA
Bulk-edit ALL checklist items of a project/portfolio (item count cannot change; omitted optional fields are reset).
Inputs: items (required), entityId (required), entityType (required), notify.
update_entity_checklist_itemC
Edit a single checklist item of a project/portfolio.
Inputs: itemId (required), entityId (required), entityType (required), text, +4 more.
update_entity_commentC
Edit a comment of a project, portfolio or goal.
Inputs: text (required), entityId (required), commentId (required), entityType (required), +3 more.
update_entity_permissionsC
Grant/revoke direct permissions on an entity. roles: AUTHOR|OWNER|CLIENT|FOLLOWER|MEMBER.
Inputs: entityId (required), entityType (required), grant, revoke.
update_filterA
Update a saved filter. "filter"/"query" fully replace the previous conditions.
Inputs: filterId (required), name, query, sorts, +4 more.
update_issue_fieldA
Rename a global issue field and/or update its dropdown values, category, order, visibility (admin). Optimistic locking: pass the current field version.
Inputs: fieldId (required), version (required), name, order, +6 more.
update_issue_field_categoryC
Rename an issue field category, change its order (admin).
Inputs: categoryId (required), name, order, version, +1 more.
update_issue_typeB
Rename an issue type (admin). version locks edits to that version.
Inputs: name (required), typeKey (required), version.
update_priorityB
Edit a priority: name, description (admin). The icon cannot be changed.
Inputs: priorityKey (required), name, version, description.
update_projectC
Update project parameters.
Inputs: projectId (required), lead, name, tags, +3 more.
update_queue_local_fieldC
Edit a local field of a queue: name, category, options, visibility.
Inputs: queueId (required), fieldKey (required), name, order, +6 more.
update_queue_macroB
Edit a queue macro: name, comment body, issue updates.
Inputs: macroId (required), queueId (required), body, name, +1 more.
update_queue_triggerC
Edit a trigger: name, actions, conditions, position (version = trigger version).
Inputs: queueId (required), version (required), triggerId (required), name, +4 more.
update_queue_versionC
Edit a queue version: name, description, dates.
Inputs: versionId (required), name, dueDate, startDate, +1 more.
update_resolutionB
Edit a resolution: name, description, order (admin).
Inputs: resolutionKey (required), name, order, version, +1 more.
update_sprintB
Rename a sprint, change dates or status (version = sprint version, If-Match lock).
Inputs: version (required), sprintId (required), name, status, +2 more.
update_statusC
Edit an issue status: name, description, order, category (admin).
Inputs: statusKey (required), name, type, order, +2 more.
update_workflowA
Update a workflow (admin). Pass only the fields to change; version is required for optimistic locking (412 on conflict).
Inputs: version (required), workflowId (required), name, type, +3 more.
update_workflow_actionB
Update one transition action of a workflow step (admin): name, target status, screen, conditions. Version is the current workflow version.
Inputs: version (required), actionId (required), statusKey (required), workflowId (required), +6 more.
upload_issue_attachmentA
Attach a local file to an issue. Max file size is 1024 Mbit. The file must exist on the machine where the MCP server runs.
Inputs: issueId (required), filePath (required), filename.
upload_temp_attachmentA
Upload a temporary file (not yet attached anywhere). Returns a temp id usable once in create_issue attachmentIds, add_comment attachmentIds and entity attachments.
Inputs: filePath (required), filename.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 187 tools
Multiple tool pairs have unclear boundaries, especially issue-vs-entity duplicates (add_comment/add_entity_comment, get_issue_attachments/list_entity_attachments), pagination variants (get_boards/get_boards_paginate, get_users/get_users_relative), field scopes (get_issue_fields/get_queue_fields/get_queue_local_fields), access-vs-permission tools (get_entity_access/get_entity_permissions, update_entity_access/update_entity_permissions), and attachment upload routes (upload_temp_attachment vs upload_issue_attachment). Descriptions help somewhat, but with 187 tools an agent will likely misselect among these closely related options.
Names are consistently snake_case and mostly verb-first (get_, create_, update_, delete_), but the set mixes synonymous verbs: get_ vs list_, create_ vs add_, and update_ vs edit_. These synonyms are semantic and readable, yet they break strict predictability across otherwise similar operations.
187 tools is far beyond the 50+ threshold for extreme mismatch and creates enormous context and selection overhead for an agent. Even though the domain is broad, the surface should be consolidated or grouped to be practically usable.
Core issue, board, project, entity, comment, attachment, link, and workflow operations are well covered, but notable lifecycle gaps remain: no delete_issue, no update_queue, no delete for triggers/autoactions, no delete for statuses/priorities/resolutions/issue types, and incomplete dashboard/report management. These gaps will create dead ends for common administrative workflows.
Maintenance
Related MCP Connectors
Task management for people and AI agents, with scoped OAuth access to issues, projects, and docs.
Task management for people and AI agents, with scoped OAuth access to issues, projects, and docs.
Task manager your agent can fully operate: boards, tasks, sprints, roles, worklogs, day planner.
Issue tracking for software projects: issues, epics, releases, docs and boards, from your AI.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceEnables AI assistants to interact with Yandex.Tracker task management system through MCP protocol. Supports creating and managing issues, searching tasks, handling comments, managing projects and queues, and generating analytics reports.ISC
- AlicenseNot gradedqualityDmaintenanceEnables interaction with Yandex Tracker through its API for managing tasks, comments, and attachments. It supports issue searching, status transitions, and metadata retrieval for automated project management.MIT
- FlicenseNot gradedqualityDmaintenanceEnables AI agents to manage Yandex Tracker tasks with tools for creating, reading, updating, and transitioning issues via a Streamable HTTP endpoint.-
- AlicenseAqualityBmaintenanceEnables AI assistants to securely interact with Yandex Tracker, managing issues, queues, comments, worklogs, and search, including attaching files to issues.561Apache 2.0