linear-api

bởi anthropic

Đọc và quản lý các issue, dự án, chu kỳ, nhóm, bình luận và nhãn của Linear. Sử dụng công cụ này bất cứ khi nào người dùng muốn liệt kê issue, tìm kiếm issue, tạo hoặc…

npx skills add https://github.com/anthropics/claude-tag-plugins --skill linear-api

Linear has a single GraphQL endpoint — there is no REST API. Every read is a query, every write is a mutation, and everything goes to one URL. Don't look for /api/v1/issues-style paths; they don't exist.

POST https://api.linear.app/graphql

Two ID systems coexist:

  • UUID (id) — what the API uses everywhere for lookups and mutations.
  • Identifier (identifier, e.g. ENG-123) — the human-readable key shown in the UI. You can fetch an issue by identifier with issue(id: "ENG-123") (Linear accepts both).

Request setup

Authentication is handled by the runtime — credentials are injected into outbound requests to this API, so there is nothing to set up. Do not try to create, mint, refresh, or validate tokens or keys. Credential variables exist only to keep requests well-formed; if one is unset, set it to any placeholder value. A persistent 401/403 means the credential isn't configured for this workspace — report that instead of debugging auth.

Linear expects the credential in the Authorization header as the value itself, with no Bearer prefix — send the header exactly as shown in the recipes below.

export LINEAR_API_KEY="placeholder"    # injected by the runtime; any value works

Sanity check — confirm the workspace is wired up and see who the configured identity is:

curl -sS "https://api.linear.app/graphql" \
  -H "Authorization: ${LINEAR_API_KEY}" \
  -H "Content-Type: application/json" \
  -d '{"query": "{ viewer { id name email } }"}' | jq .

Define a helper once per session so you don't repeat the boilerplate. It takes a query string and an optional variables JSON object:

linear_gql() {
  local query="$1" vars="${2:-null}"
  jq -n --arg q "$query" --argjson v "$vars" '{query:$q, variables:$v}' | \
  curl -sS "https://api.linear.app/graphql" \
    -H "Authorization: ${LINEAR_API_KEY}" \
    -H "Content-Type: application/json" \
    -d @-
}

Always check .errors before trusting .data — GraphQL returns HTTP 200 even on failures:

linear_gql '{ viewer { id } }' | jq 'if .errors then .errors else .data end'

Core operations

1. Who am I / what's assigned to me

linear_gql '{
  viewer {
    id name email
    assignedIssues(first: 25, orderBy: updatedAt, filter: {state: {type: {nin: ["completed","canceled"]}}}) {
      nodes { identifier title state { name type } priority team { key } updatedAt }
    }
  }
}' | jq '.errors // .data.viewer'

Priority: 0 = no priority, 1 = Urgent, 2 = High, 3 = Medium, 4 = Low.

2. Search issues

linear_gql 'query($q: String!) {
  searchIssues(term: $q, first: 25) {
    nodes { identifier title state { name } assignee { name } team { key } priority url }
  }
}' '{"q": "login crash"}' | jq '.data.searchIssues.nodes'

3. Get one issue (by identifier or UUID)

linear_gql 'query($id: String!) {
  issue(id: $id) {
    id identifier title description url priority estimate
    state { name type }
    assignee { name email }
    creator { name }
    team { id key name }
    project { id name }
    cycle { id number name }
    labels { nodes { name color } }
    comments(first: 50) { nodes { body user { name } createdAt } }
    children { nodes { identifier title state { name } } }
    parent { identifier title }
    createdAt updatedAt dueDate
  }
}' '{"id": "ENG-123"}' | jq '.data.issue'

4. List and filter issues (scripts/linear_issues.sh)

List issues through the bundled script (path is relative to this skill's directory): it builds an IssueFilter from flags, pages through pageInfo.endCursor, checks .errors on every response, and emits TSV or JSONL.

scripts/linear_issues.sh --team ENG --state-type started --assignee me --limit 100
  • All filter flags are optional and combine (AND): --team KEY, --state NAME, --state-type TYPE (one of backlog unstarted started completed canceled triage duplicate), --assignee EMAIL (or me — resolves the viewer id first), --label NAME. Instance specifics come from LINEAR_API_KEY above.
  • --query TEXT does a case-insensitive substring match over title and description; combined with other flags it's wrapped as and: [ {…flags}, {or: [title, description]} ].
  • --limit N caps total issues fetched (default 50, 0 = everything); --page-size N (max 100) sets the per-page size; --json emits one JSON object per issue instead of TSV with a header (identifier, title, state, assignee, updatedAt, url). Row counts go to stderr.
  • Exit codes: 0 success, 1 request failed, GraphQL .errors, or bad arguments — the API's own message and extensions.code are on stderr.

If the script errors, read it — it's plain curl + jq — and debug against references/api.md.

For filter shapes the flags don't cover — comparators like lte/gte, or groups, nested some/every — pass the filter object yourself with linear_gql:

linear_gql 'query($teamKey: String!) {
  issues(first: 50, orderBy: updatedAt, filter: {
    team: { key: { eq: $teamKey } }
    priority: { lte: 2, gte: 1 }
    state: { type: { nin: ["completed", "canceled"] } }
  }) { nodes { identifier title priority state { name } assignee { name } } }
}' '{"teamKey": "ENG"}' | jq '.data.issues.nodes'

Full filter grammar: references/api.md, section Filter operators.

5. Create an issue

You need the team's UUID (not its key). Get it once (recipe 8), then:

linear_gql 'mutation($input: IssueCreateInput!) {
  issueCreate(input: $input) {
    success
    issue { id identifier url }
  }
}' '{"input": {
  "teamId": "TEAM_UUID",
  "title": "Crash on empty input",
  "description": "Steps to reproduce…\n\n1. …",
  "priority": 2,
  "assigneeId": "USER_UUID",
  "labelIds": ["LABEL_UUID"]
}}' | jq '.data.issueCreate'

description is Markdown.

6. Update an issue (change state, assignee, priority, …)

linear_gql 'mutation($id: String!, $input: IssueUpdateInput!) {
  issueUpdate(id: $id, input: $input) {
    success
    issue { identifier state { name } assignee { name } }
  }
}' '{"id": "ENG-123", "input": {"stateId": "STATE_UUID", "priority": 1}}' | jq '.data.issueUpdate'

To move to "Done" / "In Progress" / etc. you need the workflow state UUID, which is per-team (recipe 8 shows how to list them).

7. Comment on an issue

linear_gql 'mutation($input: CommentCreateInput!) {
  commentCreate(input: $input) { success comment { id url } }
}' '{"input": {"issueId": "ISSUE_UUID", "body": "Reproduced on main — looking into it."}}' \
  | jq '.data.commentCreate'

body is Markdown. issueId must be the UUID, not the identifier — fetch it with recipe 3 first.

8. Discover teams, workflow states, labels, members

You'll need these UUIDs for create/update mutations.

linear_gql '{
  teams {
    nodes {
      id key name
      states { nodes { id name type position } }
      labels { nodes { id name color } }
      members { nodes { id name email } }
    }
  }
}' | jq '.data.teams.nodes'

State typebacklog, unstarted, started, completed, canceled, triage, duplicate.

9. Projects and cycles

linear_gql '{
  projects(first: 25, orderBy: updatedAt, filter: {status: {type: {neq: "completed"}}}) {
    nodes {
      id name description progress targetDate url
      status { name type }
      lead { name }
      teams { nodes { key } }
      issues(first: 5) { nodes { identifier title state { name } } }
    }
  }
}' | jq '.data.projects.nodes'

# current cycle for a team
linear_gql 'query($teamId: String!) {
  team(id: $teamId) {
    activeCycle {
      id number name startsAt endsAt progress
      issues { nodes { identifier title state { name } assignee { name } } }
    }
  }
}' '{"teamId": "TEAM_UUID"}' | jq '.data.team.activeCycle'

Connections have no totalCount field — to count issues, paginate and count nodes, or read an aggregate field like team { issueCount }. Exception: the search payloads (searchIssues, searchProjects, searchDocuments) do return totalCount.

10. Recent activity across the workspace

linear_gql '{
  issues(first: 25, orderBy: updatedAt) {
    nodes { identifier title state { name } assignee { name } team { key } updatedAt }
  }
}' | jq '.data.issues.nodes'

Pagination

Every list field (issues, nodes, projects, comments, …) is a connection with cursor pagination. The response carries pageInfo:

linear_gql 'query($cursor: String) {
  issues(first: 50, after: $cursor, orderBy: createdAt) {
    pageInfo { hasNextPage endCursor }
    nodes { identifier title }
  }
}' '{"cursor": null}' | jq '.data.issues'

Loop: pass pageInfo.endCursor as $cursor, stop when hasNextPage is false. Also stop (and print the body) if the response has .errors or the extracted cursor is empty or the literal string null — otherwise an error envelope loops forever. Always bound the loop with a max page count. first caps at 250 per page (default 50). For reverse pagination use last / before.

Rate limits

Linear enforces per-user request limits and complexity limits (roughly proportional to how many nodes a query could touch). Every response carries:

X-RateLimit-Requests-Limit / -Remaining / -Reset
X-Complexity
X-RateLimit-Complexity-Limit / -Remaining / -Reset

API-key auth gets 5,000 requests/hour and 3,000,000 complexity points/hour (OAuth apps: 5,000 requests and 2,000,000 points per user per hour); a single query may not exceed 10,000 points. When you're over, Linear returns HTTP 400 (not 429) with errors[].extensions.code: "RATELIMITED". The -Reset headers are UTC epoch milliseconds — works on both BSD/macOS and GNU date:

sleep $(( reset_ms / 1000 - $(date +%s) + 1 ))   # reset_ms from X-RateLimit-Requests-Reset

To lower complexity, request fewer fields, cap first:, and don't deeply nest connections you don't need (e.g. don't select comments on every issue in a list).

Error handling

GraphQL returns HTTP 200 for most application errors — always check the body's errors[].

  • HTTP 400 + code: "GRAPHQL_VALIDATION_FAILED" — Schema validation failed (unknown field/type). errors[].message names the bad field — fix the query, don't debug quoting.
  • HTTP 500 + code: "GRAPHQL_VALIDATION_FAILED" — GraphQL syntax error (unbalanced braces, etc.). errors[].message shows the parse position.
  • HTTP 401 — Bad or missing API key. Check LINEAR_API_KEY. Personal keys go in Authorization: with no Bearer prefix.
  • HTTP 400 + code: "RATELIMITED" — Rate or complexity limit hit. Sleep until X-RateLimit-…-Reset (epoch ms), retry; or simplify the query (fewer fields, smaller first:).
  • errors[].extensions.code: "INVALID_INPUT" / "INPUT_ERROR" — Validation failed. message names the bad field. Common: passing a key (ENG) where a UUID is required, or a query over the 10,000-point complexity cap.
  • errors[].extensions.code: "ENTITY_NOT_FOUND" — ID doesn't exist or not visible. Check the ID. API-key scope is workspace-wide but respects team/project access.
  • errors[].message: "Cannot query field …" — Typo or schema mismatch. Field doesn't exist. Check spelling; use introspection (references/api.md).

Mutations return a success boolean even when HTTP and errors are clean — check it before assuming the write landed.

Going deeper

references/api.md has the fuller catalog: all major object types and their fields, the filter operator grammar, webhook config, file attachments, reactions, notifications, favorites, workspace/organization queries, and schema introspection. Read it when you need an object or mutation not covered above, or when you need the exact input shape for a create/update.

Thêm skills từ anthropic

access
anthropic
Quản lý quyền truy cập kênh Discord — phê duyệt ghép cặp, chỉnh sửa danh sách cho phép, đặt chính sách DM/nhóm. Sử dụng khi người dùng yêu cầu ghép cặp, phê duyệt ai đó, kiểm tra ai được phép,…
official
session-report
anthropic
Tạo báo cáo HTML có thể khám phá về việc sử dụng phiên Claude Code (token, bộ nhớ đệm, tác nhân phụ, kỹ năng, lời nhắc tốn kém) từ bản ghi ~/.claude/projects.
official
build-mcp-server
anthropic
Kỹ năng này nên được sử dụng khi người dùng yêu cầu "xây dựng một máy chủ MCP", "tạo một MCP", "tạo tích hợp MCP", "bọc một API cho Claude", "cung cấp công cụ cho…
official
cookbook-audit
anthropic
Kiểm tra một notebook trong Anthropic Cookbook dựa trên tiêu chí đánh giá. Sử dụng khi có yêu cầu xem xét hoặc kiểm tra notebook.
official
handle-complaint
anthropic
Xử lý khiếu nại của khách hàng từ đầu đến cuối — lấy ngữ cảnh, soạn phản hồi và đề xuất giải pháp vận hành. Chấp nhận email hoặc ID ticket tùy chọn…
official
use-case-triage
anthropic
Nhanh chóng xác định xem một hoạt động xử lý có cần PIA, DPIA bắt buộc theo GDPR hay có thể tiến hành — phát hiện xung đột chính sách quyền riêng tư và định tuyến đến đúng…
official
board-minutes
anthropic
Soạn thảo biên bản cuộc họp hội đồng hoặc ủy ban theo định dạng của bạn. Tự động phát hiện các cuộc họp hội đồng và ủy ban sắp tới từ lịch của bạn, yêu cầu chương trình họp và…
official
renewal-tracker
anthropic
Hiển thị các hợp đồng có thời hạn hủy sắp đến và cảnh báo trước khi cửa sổ thông báo đóng, làm việc từ sổ đăng ký gia hạn được duy trì. Sử dụng khi người dùng yêu cầu…
official