Lenny Rachitsky · 9 min read · Gen AI Tools
🎙️ How I AI: GPT-5.6 review, How a solo builder runs 24/7 local AI, and What an agent harness is and how to build one
Read the original on lennysnewsletter.com →
The episode explains what a harness is—code around an AI agent to make it effective for a specific job—and shows how to build one with the Claude Agent SDK to automate Sentry bug triage. It argues for constrained, task-specific harnesses and multi-model routing over general-purpose agents, and notes how a solo builder runs 24/7 local AI.
Key Takeaways
- A harness is code around an AI agent to make it effective for a specific job.
- Build harnesses for repeatable workflows where steps/tools are defined but the AI decides the rest (e.g., Sentry bug triage).
- Use specialized adapters instead of broad tool access to improve speed, cost, and safety.
- Encode permissions in the harness to avoid repeating prompts and enforce constraints.
- Output structured artifacts (task logs, a Sentry issue brief, relevant logs, a worker report, and an HTML summary file) for a consistent, team-wide record.
Excerpt
What a harness is and how to build one with Claude Agent SDK Listen now on YouTube • Spotify • Apple Podcasts Brought to you by: Bolt.new —Turn your idea into a real product Customer.io —Build customer engagement campaigns from a single prompt Claire explains why harnesses matter and when they’re better than general-purpose tools like Claude Code or Codex, and walks through the custom Claude Agent SDK harness she built to automate Sentry bug triage at ChatPRD. You’ll see how she structured the workflow, encoded permissions, connected tools like Sentry and Linear, and turned a repeatable engineering task into…