sprint planning with Claude Code + Linear MCP
Sprint I:01-3 planning. Jan 25 to Feb 8. Instead of bouncing between Linear, Slack, and a doc, I ran the whole thing from Claude Code using the Linear MCP server.
Workflow:
- Pull the current cycle from Linear, see what’s in flight
- Review priorities with context from Slack discussions
- Create and assign tickets directly from the planning session
- Document the sprint plan as structured markdown
What the cycle looked like when I pulled it:
| ID | Title | Priority | Status | Assignee |
|---|---|---|---|---|
| ECHO-636 | Fix rate limit issues with chat (LLM Router) | Urgent | Ready for QA | Sameer |
| ECHO-637 | RangeError - Transcription fails when screen auto-dims | Urgent | Development | Usama |
| ECHO-606 | Collect human reference transcripts | Urgent | Todo | Robert |
| ECHO-634 | Design changes for product | Medium | Todo | Jorim |
| ECHO-587 | Context selection modal: Select all (+ tags) | Low | Ready for QA | Usama |
Priorities for this sprint:
P1 - Workspaces MVP. Users belong to multiple workspaces with seamless switching. URL structure: app.dembrane.com/{workspace_slug}/projects/{project_id}. Auto-workspace for existing users so nothing breaks. Explicitly NOT in MVP: inviting users, billing, partner relationships, roles.
P2 - Design & branding consistency. New brand definition rolling out.
P3 - QA automation. Bringing on a freelancer for structured QA. Budget: EUR 400, supervised by Usama.
The thing that made this work: having Linear data inline while discussing priorities eliminated context-switching. No tab switching. No copy-pasting issue IDs. Pull the data, discuss, create tickets, move on.
Where it fell short: MCP server hit auth issues occasionally (FORBIDDEN on some Directus queries), and creating tickets with all the right metadata (cycle assignment, labels, priority) required knowing the exact field names. Not a drag-and-drop board. But for a CTO doing sprint planning solo before bringing it to the team, significantly faster than the GUI.
The sprint plan became a markdown doc with P1/P2/P3 sections, explicit scope boundaries (“NOT in MVP”), and budget allocations. Generated from the same session where I created the tickets, so everything stayed consistent.
Would do this every sprint. The 30 minutes saved on context-switching compounds.