Use case

One AI agent identity across Cursor, Cline, Claude Code, and Windsurf

Senior engineers use multiple IDEs per day: Cursor for fast iteration, Cline for autonomous loops, Claude Code for long sessions, Windsurf for the Cascade agent. GenZAgents unifies all of them into one receipt feed.

The senior-engineer IDE-hop pattern

A typical senior engineer's day: Cursor in the morning for code review and inline edits (fast turnaround). Cline for a 90-minute autonomous loop on a feature spec. Claude Code for a deep refactor session. Windsurf's Cascade for the multi-file migration. Each IDE has different strengths; engineers pick the right tool per task. The problem: each IDE has its own chat history, its own context, its own cost. The receipt feed unifies them.

How one agent DID handles all four IDEs

The MCP server config (or the @genzagentsio/setup CLI) installs the same agent DID into every IDE's MCP config. Every receipt — regardless of IDE — is tagged with that DID. The dashboard sees one continuous timeline; you can filter "show me only Windsurf receipts" if you want IDE-specific views, but the default is one feed across all of them.

Project tagging unifies the context

Set GENZAGENTS_PROJECT=acme-checkout in the .env file of the project. Every receipt from any IDE working in that directory inherits the tag. So even though you used 4 different IDEs over the day, the dashboard view "show me everything tagged acme-checkout today" gives the complete picture.

Cost rollup per IDE for optimisation

The receipts log which IDE produced them (runtimeProvider field). Roll up cost by IDE: "you spent £42 in Cursor today, £18 in Cline, £6 in Windsurf, £2 in Claude Code." If the rollup shows you're over-investing in one IDE without proportional output, that's a tuning signal.

Cross-IDE handoff in practice

You're in Cursor finishing a refactor; you realise the autonomous-loop pattern from Cline would handle the next step better. Save the context (auto-receipt-on-commit fires), open Cline on the same repo, the MCP-injected restore_chat surfaces the Cursor session as a candidate. Click restore; Cline picks up where Cursor left off. No re-briefing.

Why this matters strategically

IDE switching is high-leverage for senior engineers — pick the right tool per task. But it's currently friction-heavy because the context doesn't travel. GenZAgents removes the friction without forcing IDE consolidation. Engineers keep using all 4 IDEs; the receipt feed makes the cross-IDE pattern feel native.

Common questions

Do I need separate API keys per IDE?

No — one API key (in ~/.genzagents-mcp-env, mode 600) is read by the MCP server regardless of which IDE invoked it. Configure once, works everywhere.

What if I want different agent identities for different IDEs?

You can — set GENZAGENTS_AGENT_DID=<did> per-IDE in the IDE-specific MCP config. Useful for splitting agent identities by workflow class (e.g. autonomous-loop agent for Cline, interactive agent for Cursor).

Does the unified receipt feed slow down switching?

No — each IDE's MCP server runs independently in its own process. There's no cross-process locking; receipts flow to the central API in parallel.

What about VS Code with Continue.dev — same flow?

Yes — Continue.dev with the GenZAgents MCP config is treated as another IDE. Receipts unify the same way.

Related

Get the trust layer for your AI work

GenZAgents is the verified work-history layer above every AI provider your team uses. Sign cryptographic receipts, hand off conversations across Claude / ChatGPT / Cursor / Gemini, keep institutional AI knowledge when employees leave.

Last reviewed · 3 min read· Open spec· Changelog