# How do I set up agents that do real work?

## The short answer

- Rob Hallam: Route Claude Code work: cheap scout reads, top model plans and builds, fresh reviewer diffs

## Playbooks

### [Route Claude Code work: cheap scout reads, top model plans and builds, fresh reviewer diffs](https://vibedraft.app/builder-brains/rob-hallam/route-claude-code-work)

Rob Hallam · Building · X posts

Per task: scout agent (Opus, low effort, read-only) explores; the top model plans (high effort) then implements (medium) only files in a 'Files touched' list and writes an audit; Opus reviews the diff in a fresh context with Blocking/Non-blocking/Verdict output. Agent files live in ~/.claude/agents/, orchestrated by CLAUDE.md.

**When it applies:** Claude Code users hitting weekly usage limits

Source: [Rob Hallam: output-optimized Fable stack for Claude Code (thread)](https://x.com/robj3d3/status/2072603182149386299), 2 July 2026 on x.com

## Sources worth reading

- [Multi-agent systems are an absolute nightmare in production](https://vibedraft.app/builder-brains/sources/4e5d7a16-08f2-46e9-b451-7746152b737c): Practitioner rant on multi-agent systems in production: retries, recovery and why they break.
- [Claude improved my agent harness by 40.7% overnight](https://vibedraft.app/builder-brains/sources/a23b6027-fa8e-4f5c-a98c-8184c0688b19): First-hand: using Claude to improve an agent harness by 40.7% overnight, an autoresearch-style loop.
- [WTF Is an "AI Agent Loop"? Genius or Hype?](https://vibedraft.app/builder-brains/sources/2fd8fa20-0411-4699-9f57-18e98dc4442a): What an agent loop actually is and when it beats a single prompt, explained with examples.
- [The agent harness belongs outside the sandbox](https://vibedraft.app/builder-brains/sources/c45d1f3d-0b5e-4c4c-9b8b-2112bac9d228): Argues the agent loop should run outside the sandbox and send only risky tool calls in; HN commenters push back.
- [$100M ARR Founder: Build an Agent Harness](https://vibedraft.app/builder-brains/sources/58034184-c22e-4fba-84db-dbf83b87cb2c): A $100M ARR founder on building an agent harness as the product's core.

## Founders who talk about it

- [Rob Hallam](https://vibedraft.app/builder-brains/rob-hallam)

## Related topics

- [What can I automate with AI as a solo founder?](https://vibedraft.app/builder-brains/topics/ai-automation)
- [How do I keep AI costs from eating my margin?](https://vibedraft.app/builder-brains/topics/ai-costs)
- [Will AI make my startup pointless?](https://vibedraft.app/builder-brains/topics/ai-platform-risk)
- [How do I build an AI product that isn't just a wrapper?](https://vibedraft.app/builder-brains/topics/ai-products)
- [How do I ship faster with Claude Code and Cursor?](https://vibedraft.app/builder-brains/topics/coding-agents)
- [How do I write prompts and context that work?](https://vibedraft.app/builder-brains/topics/context-engineering)

Updated 2026-11-07. [Web version](https://vibedraft.app/builder-brains/topics/agent-workflows)

Advice cards are rewritten summaries with a link to the original. They are not the founder's words unless shown in quotation marks.
