New MCP Server

Most MCP servers look up a ticket. DevStride delivers the work.

Point your agent at a typical project-tool MCP and it looks up a ticket. Point it at DevStride and it assesses risk, drafts the change, plans the whole build, then applies it the moment you say go. Same protocol. A different class of partner.

Works with Claude, ChatGPT, Cursor, Copilot, Codex, and Gemini CLI. Connect in minutes from the Connect AI page. Your AI tool's own subscription is separate. Read the MCP docs →
Connects to the AI you already use
However you deliver

The same loop, whatever the work.

Whether you ship a product or deliver for clients, the pattern is identical: the MCP Server does the legwork, you make the calls.

Services & client delivery
  • Scope a client engagement into a plan the whole team can see.
  • Spot an at-risk milestone and reallocate to protect the date.
  • Turn out a polished, client-ready status update every week.
Product delivery
  • Draft the plan, then run the delivery loop story by story.
  • Surface what's slipping and rebalance to hit the release.
  • Draft the sprint status for stakeholders, ready to review and send.
The capability leap

Beyond read-only. Beyond light edits.

Most MCP integrations stop at read and light edits. DevStride goes a level further, and the judgment stays with you.

Read

An agent can see your data and answer questions about it.

Most tools
Edit

An agent can nudge a field, change a status, or post an update.

Some tools
Partner

Assesses risk, drafts the plan, applies the change you approve. The work a delivery lead does, with you still making the call.

DevStride

It does the work a delivery lead would hand off. You stay the delivery lead.

A true working partner

It proposes. You approve. It applies.

Your agent works with your permissions. It proposes, you approve, it applies. Most MCP integrations stop at read and light edits; DevStride's assesses risk, proposes changes, even drafts whole plans.

Replace · real capture
claude · devstride-mcp
How is delivery throughput trending this quarter?
# calling get_throughput
Throughput is up quarter over quarter. Three teams ahead of plan, and the Checkout program has capacity to pull work forward.
Pull the next two Checkout items into this cycle and update the affected dates.
# drafting proposal · nothing applied yet
Proposed: move 2 items into the current cycle, update the affected dates. ⏸ Waiting on your go-ahead.
Approved.
✓ Applied with bulk_update_items. Every change attributed to you.
Draft a status report for leadership on this quarter's gains.
# get_velocity → drafting summary
Draft ready: the quarter's throughput gains, teams ahead of plan, Checkout pulled forward. Yours to review and send.

Assess

AI pulls the portfolio, traces dependency chains, and surfaces what's trending late.

You decide what actually matters.

Modify

AI drafts a rebalance and shows which items move and who picks up the load.

You review the proposal and give the go-ahead.

Create

AI drafts a full plan: parent items, child work, boards, and the dependency stack.

You vet and refine the plan.

Built around your judgment

Plan with AI. Deliver on a loop. Keep the judgment.

AI does the smart grunt work. People make the judgment calls.
How we use it ourselves

We planned and shipped our own Service Desk through this MCP, every item tied to the change that built it.

Using the DevStride MCP Server, our team had an agent draft the entire build: the parent item, the child work, the boards, and the full dependency stack. We vetted the plan, then execution backfilled against it.

Every item links to the change that implemented it, in order. Someone with zero context can see exactly what happened, why, and when. That's the organizational memory you get when your AI partner works inside the all-in-one delivery platform where humans and AI work together.

Replace · real capture
EPIC Service DeskDrafted by agent, vetted by the team
CAP Ticket intake & routing3 child items · 2 owners PR #1042
CAP SLA & escalation rules4 child items · 1 owner PR #1051
DEP Notification serviceLinked dependency PR #1067
A breeze to implement

Heavy hitter. Light lift.

No glue code, no infrastructure to stand up. The MCP Server is built on the same API that powers the app, so your assistant can reach nearly everything you can do in the product. Connect in minutes from the Connect AI page.

Remote endpoint
https://api.devstride.com/mcp
01

Open Connect AI

Start from the Connect AI page in the DevStride app and pick the client you already use.

02

Sign in securely

Authorize with OAuth 2.1. The connection works with your DevStride permissions, so treat it like your own credentials.

03

Start working

Ask your agent to assess, propose, or draft. You approve from there.

Works with Claude, ChatGPT, Cursor, Copilot, Codex, and Gemini CLI, plus any MCP-compatible client. Your AI tool's subscription (Claude, Claude Code, etc.) is billed separately by that provider.

It does the heavy lifting. You make the calls.

Field client questions, scope engagements, surface insights, and turn out status reports, with the MCP Server doing the legwork and your team making every call.
Every plan includes every capability. No tiers, no surprises, and you're up and running fast.