← back
Building your own software factory — Eric Zakariasson, Cursor
Takeaway
Treat your codebase like a factory: structure for discoverability, add guardrails reactively, and invest in verifiable end-to-end checks.
Summary
- Eric Zakariasson (Cursor DX/product) describes building a 'software factory' using Dan Shapiro's six levels of autonomy, where teams progress from spicy autocomplete to a 'dark factory' of agents shipping autonomously.
- Primitives & patterns: modular, colocated codebases so agents can ls a folder and discover everything; consistent boilerplate (auth, startup scripts, tests) for agents to reference.
- Guardrails: hooks that block touching sensitive code (encryption, auth); cursor rules should emerge dynamically when agents go off-rails, not be installed wholesale from cursor.directory.
- Enablers: skills + MCPs for context, feature-flag skill so autonomous PRs ship behind flags for easy revert, dev-environment startup automation, and verifiable systems (unit, integration, UI/DOM tests).
- Argues backend systems with clear contracts are easier to give to agents than UI; the checklist is runnable, accessible (linear/notion/datadog/slack), verifiable.
cursoragentscode-generation
Original description
Most of us are pair-programming with one agent and stopping there. There's a lot more on the table. This workshop is about going from one agent to many. We'll start with codebase setup, the foundational work that makes agents effective on their own. Then we'll scale up to running agents in parallel, kicking off async work that keeps going while you context-switch to something else, and setting up automations for the things you're still doing by hand. Speaker info: - https://x.com/ericzakariasson - https://www.linkedin.com/in/ericzakariasson/ - https://github.com/ericzakariasson Timestamps: 0:00:00 - Workshop introduction and vision for autonomous software factories 0:01:26 - Frameworks and stages of agentic autonomy 0:08:58 - Establishing scalable, reproducible dev environments for agents 0:10:00 - Importance of verifiable systems and automated testing pipelines 0:10:47 - Cursor 3 walkthrough: Redesigned agent-first interface 0:15:26 - Cloud agents: Scaling via dedicated VMs and computer control 0:19:17 - Managing asynchronous workflows and frontloading context 0:24:45 - Automating repetitive tasks to build feedback loops 0:29:28 - Continual learning: Extracting rules from chat transcripts 0:30:48 - Scaling management: Moving to nested agent orchestration 0:31:08 - Strategic takeaways: Human accountability and observability 0:33:39 - Q&A: Addressing code quality and architectural guardrails 0:53:07 - Best practices for human-to-agent collaboration and handoffs 1:03:05 - Maintaining the factory: Managing documentation and specs 1:10:06 - Integrating Linear and Slack for automated issue triage 1:14:14 - Local execution: Leveraging Cursor harness locally