--- title: "Linear AI agent connections and job ideas" canonical: "https://agentuse.io/ai-agents/tools/linear" last_tested: "September 2026" --- # Linear AI agent connections and job ideas Run a Linear AI agent for the recurring coordination work around a product backlog, without handing your agent definition or runtime to a hosted workflow builder. Canonical HTML guide: https://agentuse.io/ai-agents/tools/linear ## How to connect MCP, CLI, API, and SDK options provide capabilities. A skill provides operating guidance only; it does not authenticate to or grant access to the service. ### Linear MCP Server (MCP) - Source: Official - Documentation: https://linear.app/docs/mcp - Best for: Issue, project, comment, initiative, and milestone workflows - Capability: Hosted Streamable HTTP tools for finding, creating, and updating Linear objects; a separate endpoint exposes read tools only. - Authentication: OAuth 2.1, bearer token, or a team-restricted Linear API key - Recommendation: Preferred starting point - Setup reference: Read-only: https://mcp.linear.app/mcp/readonly ยท Read/write: https://mcp.linear.app/mcp - Operating note: Use the read-only endpoint for reporting and review agents that do not need to mutate Linear. ### Linear TypeScript SDK (SDK) - Source: Official - Documentation: https://linear.app/developers/sdk - Best for: A typed, project-owned Linear connector - Capability: Strongly typed queries and mutations over Linear's public GraphQL schema. - Authentication: Personal API key or OAuth 2.0 - Setup reference: npm install @linear/sdk ### Linear GraphQL API (API) - Source: Official - Documentation: https://linear.app/developers/graphql - Best for: Custom queries, mutations, and non-TypeScript integrations - Capability: Read and mutate Linear entities through the same public GraphQL API used by Linear's applications. - Authentication: OAuth 2.0 or a permission- and team-restricted API key ### Linear Webhooks (API) - Source: Official - Documentation: https://linear.app/developers/webhooks - Best for: Event-driven starts instead of scheduled polling - Capability: Signed change events for issues, comments, projects, cycles, documents, and other supported resources. - Authentication: Webhook signing secret; admin permission is required to create workspace webhooks - Operating note: Verify the signature and timestamp, then invoke the AgentUse HTTP trigger with the validated event. ## Agent job ideas ### New-issue triage Find duplicates, missing acceptance criteria, likely team ownership, and the evidence needed before prioritization. - Services: [Linear](https://agentuse.io/ai-agents/tools/linear.md) + [GitHub](https://agentuse.io/ai-agents/tools/github.md) + [Slack](https://agentuse.io/ai-agents/tools/slack.md) - Expected output: Triage proposal ### Stale-work review Surface blocked or inactive issues whose underlying repository or customer context has changed. - Services: [Linear](https://agentuse.io/ai-agents/tools/linear.md) + [GitHub](https://agentuse.io/ai-agents/tools/github.md) - Expected output: Action list ### Release readiness brief Combine issue state with repository signals to explain what is ready, risky, or missing. - Services: [Linear](https://agentuse.io/ai-agents/tools/linear.md) + [GitHub](https://agentuse.io/ai-agents/tools/github.md) + [Slack](https://agentuse.io/ai-agents/tools/slack.md) - Expected output: Readiness brief ### Feedback-to-backlog review Group evidence-backed customer problems, recover product context, show current Linear coverage, and prepare exact issue or priority changes for review. - Services: [Linear](https://agentuse.io/ai-agents/tools/linear.md) + [Slack](https://agentuse.io/ai-agents/tools/slack.md) + [Notion](https://agentuse.io/ai-agents/tools/notion.md) - Expected output: Backlog review artifact - Full tutorial: https://agentuse.io/ai-agents/workflows/turn-product-feedback-into-a-prioritized-backlog ### Sprint-risk monitor Detect scope, dependency, and implementation signals that put the current cycle at risk. - Services: [Linear](https://agentuse.io/ai-agents/tools/linear.md) + [GitHub](https://agentuse.io/ai-agents/tools/github.md) - Expected output: Risk brief ### Support triage and escalation review Reconcile each support request with current issue coverage, then prepare a grounded reply and accountable escalation path for review. - Services: [Linear](https://agentuse.io/ai-agents/tools/linear.md) + [Gmail](https://agentuse.io/ai-agents/tools/gmail.md) + [Notion](https://agentuse.io/ai-agents/tools/notion.md) - Expected output: Support review queue - Full tutorial: https://agentuse.io/ai-agents/workflows/triage-customer-support-email-and-draft-replies ### Production-error impact review Reconcile ranked Sentry errors with existing Linear coverage, then prepare evidence-backed issue drafts without changing the backlog automatically. - Services: [Linear](https://agentuse.io/ai-agents/tools/linear.md) + [Sentry](https://agentuse.io/ai-agents/tools/sentry.md) - Expected output: Error review artifact - Full tutorial: https://agentuse.io/ai-agents/workflows/review-production-errors-by-customer-impact ### Incident action reconciliation Compare a resolved incident's evidence-backed action proposals with existing engineering work before owners create or change issues. - Services: [Linear](https://agentuse.io/ai-agents/tools/linear.md) + [Sentry](https://agentuse.io/ai-agents/tools/sentry.md) + [GitHub](https://agentuse.io/ai-agents/tools/github.md) + [Slack](https://agentuse.io/ai-agents/tools/slack.md) - Expected output: Action coverage - Full tutorial: https://agentuse.io/ai-agents/workflows/prepare-evidence-backed-incident-postmortems ## Implementation guidance for coding agents 1. Inspect the project for its installed or pinned AgentUse version and existing `.agentuse` files. 2. Load the version-matched AgentUse core, creator, and tester guidance before authoring an agent. 3. Verify which MCP servers, CLIs, APIs, SDKs, credentials, and resource IDs actually exist. Do not infer capabilities from this page. 4. Choose the smallest connection and permission set that can complete the job. Prefer read-only access for research and reporting. 5. Mechanically approval-gate consequential writes such as posting, sending, deleting, merging, or changing external records. 6. Test manually before adding a schedule or verified webhook trigger. ## Operating boundary Safe automatic work: Read scoped issues and projects, compare evidence, identify duplicates or stale work, and prepare a triage proposal. Never without the configured approval path: Create, assign, close, reprioritize, or bulk-edit issues without the configured approval path and exact target scope.