| name | cursor-tandem |
| description | User uses both Cursor and Claude Code and asks how to use them together effectively; user mentions switching between IDE and terminal AI; user want... |
Cursor + Claude Code Tandem Workflow
When to activate
User uses both Cursor and Claude Code and asks how to use them together effectively; user mentions switching between IDE and terminal AI; user wants to know which tool to use for a given task when both are available.
When NOT to use
User only has one of the two tools; user is asking about one tool in isolation without reference to the other; user wants a comparison to decide which tool to buy.
Instructions
Tool roles — keep them distinct:
- Cursor = intelligent IDE. Inline autocomplete, multi-file chat, codebase search, quick edits, component writing, reviewing diffs, exploring unfamiliar code.
- Claude Code = autonomous terminal agent. Runs shell commands, orchestrates subagents, handles multi-step tasks across many files, makes commits, sets up infrastructure.
Task routing — which tool gets the job:
Good Cursor tasks:
- Writing new components or functions
- Reviewing a diff before committing
- Exploring an unfamiliar codebase to understand structure
- Quick renames and local refactors
- Inline documentation
Good Claude Code tasks:
- Running the full test suite, then fixing failures
- Large refactors across 20+ files
- Setting up GitHub Actions, Dockerfiles, or CI/CD configs
- Database migrations
- Anything requiring bash commands or subagent orchestration
- End-to-end feature generation from spec to PR
Shared context via CLAUDE.md:
Both tools read CLAUDE.md. Write your conventions, naming rules, architecture decisions, and preferences there once — both tools respect them automatically. This is the single most important integration point.
Critical rule — never let both edit the same file simultaneously. This causes git conflicts that neither tool can resolve cleanly. Finish the Claude Code task, commit, then open in Cursor.
Handoff pattern:
- Claude Code runs the multi-step task → commits the result
- You open the commit in Cursor for fine-tuning, code review, or polish
- Cursor edits go in a follow-up commit
Parallel use pattern:
Run Claude Code in background on a long task (test suite, migration, build) while you work in Cursor on unrelated files. Claude Code reports back when done without blocking your editor workflow.
Example
"I use Cursor for writing React components and exploring the codebase. I switch to Claude Code terminal when I need to: run the full test suite, refactor across 30 files, set up GitHub Actions, or do a database migration. CLAUDE.md holds our shared conventions — both tools pick them up automatically without any extra configuration."
