MCP Server – Overview & Use Cases

• Use the MCP server to ask natural-language questions and generate reports from LinearB metrics.
• Explore persona-based use cases for leaders, managers, and executives.
• Try example prompts for releases, urgent PRs, lead time, merge rate, AI impact, and Catalog widgets.
• Use MCP tools to query metrics, repositories, linked contributor identities, releases, incidents, services, users, teams, branches, pull requests, issues, project management entities, integrations, and settings.
• Always verify outputs in the LinearB Dashboard before using results for decisions or publishing.

What to know

• Because results are generated through natural language, responses may be incomplete or inaccurate in edge cases.

As a leader, you need more than a performance snapshot. You also need context. Scaling success across squads requires visibility into the practices that drive excellence and inefficiencies that are holding you back.

With the MCP server, you can benchmark performance, identify what enables top contributors, and replicate those strategies across the organization.

Applications

• Benchmark delivery, quality, and throughput metrics across teams to spot strengths and gaps.
• Identify best practices from top contributors and scale them across the org.
• Track consistency to detect workflows or repositories that lag consistently, signaling systematic inefficiencies.
• Align internal metrics with industry benchmarks.

Coaching isn’t about instincts; it’s about data-driven insights that make growth conversations meaningful. You need to know where individuals excel and where targeted support matters.

LinearB’s MCP server surfaces contributor-level patterns in productivity, quality, and collaboration. With proper context, you can guide one-on-one conversations and drive continuous improvement.

Applications

• Spot productivity patterns around coding frequency or time allocation, and drill into specifics.
• Surface quality insights by tracking review participation and the balance of new code vs. rework.
• Reveal efficiency gaps by breaking down coding, reviewing, and deployment times.
• Provide actionable feedback using precise data to move coaching from subjective to evidence-based.

Executives and cross-functional partners often want answers, not dashboards. The LinearB MCP server makes engineering data accessible by letting you ask questions in natural language to generate tailored reports instantly.

Applications

• Deliver metrics that matter most, from merge rate to lead time, to empower executives with usable data.
• Run on-demand queries (for example, “Show me cycle time trends for the frontend repo over the last six months”).
• Filter flexibly by teams, contributors, repositories, and timeframes.

These use cases only scratch the surface. With the MCP server, you can turn raw engineering activity into actionable insight. Use reports and benchmarks to investigate trends, support coaching conversations, and communicate engineering outcomes.

Try these example questions in your connected MCP client. Replace the team, release, and reporting period with your own:

• “List the pull requests included in release [release name].”
• “Show Team Alpha’s open pull requests that need attention, including their age and review status.”
• “Create a lead time report for Team Alpha’s issues completed last month.”
• “Show Team Alpha’s merge rate for last month.”

Find and run charts and KPIs from LinearB Catalog dashboards in your MCP-compatible client.

• find_widgets finds widgets relevant to a question and returns their names, dashboards, descriptions, and chart types.
• run_widget runs a selected widget with an optional start and end date and returns chart data or a single KPI value.

Externally reported metrics submitted through the LinearB API are a separate capability: discover them with get_reported_metrics and query them with get_metrics.

Example prompts

• “Find a Catalog widget that shows rework rate.”
• “Run the Cost Per Effective PR Catalog widget from July 1 through September 30, 2026.”
• “Show the Rework Rate and PR Readiness Catalog widgets for the last 60 days.”

See Agentic Studio and the MCP Server for more information.

Use get_metrics to explore AI usage and compare engineering outcomes for AI-assisted and manual work. Supported breakdowns include coding assistants, AI code review tools, and agentic PR tools.

Distinguish AI involvement from work performed by AI

• AI-involved work: Branches where AI participated at one or more stages.
• AI-performed work: Work directly performed by AI, such as AI-written code or AI-created pull requests.

An AI review does not by itself mean the code was written by AI. Specify which definition you want when comparing results.

Available analysis

• Filter by AI-authored or co-authored commits, AI-created branches or PRs, AI reviews, interactions, and detected AI rules files.
• Explore active AI users, AI actions, and AI lines accepted, added, and deleted.
• Review GitHub Copilot usage, suggestions, acceptances, and lines, or Cursor activity such as tab, agent, chat, Cmd+K, and Composer requests.
• Analyze gitStream AI reviews, accepted suggestions, and detected or resolved issues involving bugs, security, performance, readability, maintainability, and scope.
• Use list_dev_tools to find AI tool names and identifiers, including names corresponding to IDs returned by metrics queries.

Example prompts

• “What percentage of our code changes last month were written directly by AI?”
• “Compare cycle time for AI-assisted versus manual work over the last quarter, grouped by coding assistant. Explain how AI-assisted work is defined.”
• “How many active Copilot and Cursor users did we have each week this quarter?”
• “Which AI code review tools identified the most bugs and security issues this month, based on available data?”

Metrics and benchmarks

Query metrics such as Time to Approve, Time to Merge, PRs Merged Without Review, PR Maturity, Review Depth, Active Git Days, Coding Days, and new code, refactor, and rework lines. Specify a supported aggregation such as average, p50, p75, p90, p95, or p99. If you omit it, the server uses your organization's “calculate metrics by” setting.

Group results by supported dimensions such as team, repository, user, or PR label, and roll up results by day, week, or month. Team results include parent-group information; ask your client to account for the hierarchy when combining results.

Use get_reported_metrics to discover externally reported metrics, then query their names with get_metrics using supported aggregations. Use get_db_schema to understand Metrics Builder views, columns, metric definitions, and derived calculations.

• “What custom metrics do we report to LinearB through the API?”
• “What does Review Depth measure, and how is it calculated?”
• “How does our p75 cycle time compare with the Elite, Strong, Fair, and Needs Focus benchmark tiers?”

Planning and sprints

Discover Jira boards, team iterations, and project settings. Filter issues by board, epic, project, label, issue type, or custom fields such as priority and severity. get_issues returns up to 500 issues per query; narrow filters when reviewing larger sets.

• “List Team Alpha's last three sprints and their dates and states.”
• “Show high-priority bugs on the PLAT board that are still in progress.”
• “Which boards and sprint inclusion rules are configured for Team Alpha?”

Integration health and incidents

Inspect integration connection status and the incident definitions used for change failure rate (CFR) and mean time to restore (MTTR). Incident searches include API-reported incidents and Jira or Azure Boards bugs classified as incidents by your organization's CFR settings, with each result identifying its source.

• “Which of our integrations are currently failing?”
• “How is change failure rate configured for our organization?”
• “List incidents for Team Alpha last month and identify their sources.”

The LinearB MCP server exposes tools that let you interact with your organization’s data through Claude or other MCP-compatible clients. The 25 tools below support data discovery, queries, filtering, and reporting. Use get_tool_documentation for detailed tool usage and supported parameters.

The MCP server provides dynamic access to LinearB’s data through natural language. However, as with any generative AI system, it may occasionally return inaccurate or incomplete results.

To ensure accuracy, always verify numbers and metrics in the LinearB Dashboard before making key decisions or publishing derived reports.

Agentic Studio and the MCP Server

Configuring the MCP Server