If we already have an MCP server, can we connect LinearB to the existing one?

Yes, you can use the LinearB MCP alongside any other MCP server on your preferred client, including Claude Desktop and ChatGPT.

Does the LinearB MCP only get data from Git, or can it check Jira too?

Yes, Lead Time is supported based on project management tool data. LinearB is actively working on expanding coverage of additional project management information.

How can I use the MCP to output a consistent dashboard?

You can create custom agents to build and deliver reports to your Slack using the LinearB MCP. A library of pre-made prompts is also available to get started quickly. MCP-generated numbers should be verified in the LinearB UI dashboard before use in critical decisions.

Does role-based access control apply to the LinearB MCP?

Permissions-scoped access for MCP is not currently available but is actively being developed and is on the LinearB roadmap.

Does LinearB's MCP work with GitLab, GitHub, Bitbucket, and Jira?

Yes, all Git providers and project management tools supported by LinearB are MCP-enabled.

The context layer for your software factory

LinearB connects every tool in your SDLC into one record, from ticket to release, so your agents and leaders work from the same context.

Context for your software factory Sample data

Collect
Normalize
Enrich
Augment
Serve

Collect

Every commit, ticket and build across your SDLC

LinearB connects to the tools your teams already use, from planning, code and CI/CD to AI tools like Copilot, Claude Code and Cursor. Incidents and deployments come in through its API.

Normalize

Every repo and board, in one view

Pull requests from GitHub, GitLab, Bitbucket and Azure Repos sit in one list, and issues from Jira and Azure Boards in another, so every team shows up in the same views.

Enrich

People and PRs, linked across tools

Each person's accounts merge into one contributor, every PR links to its ticket and release, and AI-assisted PRs are flagged so you can compare them with the rest.

Augment

Metrics benchmarked against 8.1M+ PRs

LinearB adds metrics like cycle time, pickup time and rework to the linked records, benchmarked against 4,800+ teams and tiered from Elite to Needs focus.

Serve

Answers for people, context for agents

People ask the LinearB AI assistant and follow dashboards and Slack, while agents read the same context through the MCP server and gitStream acts on it in your PRs.

Incoming context Live 657 events this hour

PR #9185 opened in payments-api AI-assistedGitHub
PLAT-311 moved to In Review Jira
1.2M tokens of Claude Code use in payments-api Claude Code
Build #2231 passed on payments-api Jenkins
MR !412 merged into main GitLab
38 suggestions accepted in checkout-web Copilot
Work item 5521 set to Active Azure Boards
Pipeline #884 passed for v4.12 CircleCI

Every tool, as it happens, from 50+ integrations

pull_request #9185
merge_request !412
pullrequest 77
pullRequest 318
issue PLAT-311
workitem 5521

Platform team

Pull requests

#9185 Retry refund webhooks In review
!412 Cache terraform plans in CI Merged
#77 Rotate payment API keys Open
#318 Alerting for replica lag Merged

Issues

PLAT-311 Webhook refunds time out In review
5521 Cost tags on RDS instances Active

Every repo and board in one view

Contributor Platform team

MC Maya Chen Three accounts, one contributor

@mchen
maya.c
Maya Chen

PR #9185 AI-assisted payments-api

PLAT-311 Webhook refunds time out Issue in Jira
PR #9185 Retry refund webhooks Opened by Maya Chen
v4.12 Released to production Two days after the PR opened

Every PR linked to its ticket, its release and the person behind it

Team benchmarks Platform team · last 30 days

Cycle time 1.8 days Good
Pickup time 7.1h Fair
Review time 9h Good
Deploy frequency 0.6 a day Good
Rework rate 12.3% Needs focus

Benchmarked against 8.1M+ PRs from 4,800+ teams

People ask

AI Assistant

Why is review slow on Platform?

PRs wait 7.1h for a first review, up from 3.4h. Two of five reviewers carry most of them.

Leaders track

#platform-leads

LinearB APP 9:02 AM

Pickup time on Platform is up to 7.1h this week. 14 PRs are waiting for a first review.

Agents act

Claude Code payments-api

Merge PR #9185 once the checks pass

LinearB MCP: context for PR #9185

High risk: changes refund logic684 lines across 9 filesReviewer: Platform on-call, requested by gitStream

This PR needs a human review first. I'll hold the merge until Platform on-call approves it.

The same context, wherever people and agents work

Collect Every commit, ticket and build across your SDLC50+ integrations

Normalize Every repo and board, in one view4 git providers → one PR list2 planning tools → one issue list

Enrich People and PRs, linked across toolsMaya Chen, three accountsPLAT-311 → PR #9185 → v4.12AI-assisted

Augment Metrics benchmarked against 8.1M+ PRsCycle time 1.8 daysGoodPickup 7.1hFairRework 12.3%Needs focus

Serve Answers for people, context for agentsAI assistantDashboards and SlackMCP and gitStream

AI adoption is a signal, not a result

AI speeds up coding, but the work then queues up in review and release. Adoption climbs while delivery stays flat, so track what reaches production.

Pull requests accepted

AI-generated 32.7%

Written by people 84.4%

2026 benchmarks, 8.1M+ PRs Get the report

AI Analytics Example org

What gets reported

AI-assisted PRs

41% of merged PRs, up from 14%

AI-assisted share of merged PRs

Up: AI adoption

What's happening

Cycle time by stage

44h per PR, flat since February

CodingPickupReviewDeploy

Up: Review timeDown: Coding time

A north star metric for AI ROI

Cost per effective PR adds your people cost to your AI spend and divides it by your effective PRs: merged PRs, discounted for rework. More AI spend has to show up as more delivered work.

Cost per effective PR

AI adoption How much of the work uses AI
Yield What merges without rework
Quality What holds up after release

AI ROI LinearB Catalog Example org

Cost per effective PR $602 ▼ 9% since Q2

AI spend $74k A month

Effective PRs 2,134 A month, up 11%

AI cost by toolCopilotCursorClaude Code

Cost per effective PR = (people cost + AI spend) ÷ effective PRs

Customers

How engineering leaders are proving AI impact

Gartner® Magic Quadrant™

LinearB is a Leader

Matt CulverEngineering Chief of Staff, Super.comHow LinearB closes the AI delivery gap

Renan KosickiCTO, SkeeloRead the Skeelo story

Renan KosickiCTO, SkeeloRead the Skeelo story

Context layer FAQs

A context layer brings the signals from across your SDLC, from source control, issues, CI and deploys to your AI coding tools, into one record you can trace from ticket to release. Your agents and your leaders then work from the same data.

Git providers including GitHub, GitLab, Bitbucket and Azure Repos, planning tools including Jira and Azure Boards, CI/CD tools such as CircleCI and Jenkins, and AI coding tools including GitHub Copilot, Cursor and Claude Code. Incidents and deployments can come in through LinearB's API. See all 50+ integrations.

It looks for AI co-author trailers in commits, AI agents named in commit messages or opening pull requests, and usage data from tools like GitHub Copilot, Cursor and Claude, measured against a threshold you can set.

Your total engineering cost, people and AI spend together, divided by your effective PRs, which are merged PRs discounted for rework. It's on LinearB's AI ROI dashboard, which takes your monthly AI spend and your developers' annual cost as inputs.

Yes. The LinearB MCP server brings your LinearB data into AI tools including Claude Code, Cursor, Codex, VS Code and Windsurf. gitStream acts on the same context in your pull requests on GitHub and GitLab.

More than 8.1 million pull requests from 4,800+ teams in 42 countries, published in the 2026 Software Engineering Benchmarks Report. Each metric is tiered Elite, Good, Fair or Needs focus, from the top 10% of organizations to the bottom 40%.