Integrations › Cursor
Trusted State for Cursor: project activity captured; conversation detail is limited vs Claude Code.
Get API Key →Cursor's AI reads your codebase fresh each session. Static rules help, but they can't capture dynamic knowledge — what was tried, what failed, what the team decided last week — or hand off cleanly to Claude Code.
Static rules vs. live Trusted State.
Static file vs. living memory. No contest.
One command — recommended:
npx @verifiedstate/sync init
# Configures Cursor MCP + rules + hot-context + daemon.
# Capture truth: chat/composer bubble text + activity
# (not full Claude-grade tool graphs)
# Best cross-agent pair today: Claude Code + Cursor
# Ask the second agent: "What are we working on right now?"
# Advanced: Cursor Settings → MCP → Add Server
# URL: https://mcp.verifiedstate.ai/mcp{
"receipt_id": "rct_7f3a...",
"assertion": "User prefers dark mode",
"status": "verified",
"confidence": 0.94,
"signature": "ed25519:a8c3f1...",
"verified_at": "2026-04-06T14:32:01Z",
"chain_position": 42
}| Feature | Without | With VerifiedState |
|---|---|---|
| Session context | Gone after close | Persisted + queryable History/Now |
| Cross-agent handoff | Paste a recap into Claude Code | Same workspace state via MCP + hot-context |
| Team decisions | In Slack threads | Verified assertions with timestamps |
| Capture fidelity | Nothing persisted | Activity + limited transcript (not Claude-grade) |
It complements it. vsync can manage VerifiedState rules/hot-context. Capture quality for Cursor chats is more limited than Claude Code JSONL.
Run npx @verifiedstate/sync init in your project. Or add the MCP server in Cursor settings under Other installation options.
No. Be honest: Cursor capture is activity + limited conversation text. Claude Code is the richest capture source today.
Add VerifiedState once. Keep project state available across your AI agents.