Guides

Guides for running AI coding agents

Practical guides on working with AI coding agents in parallel — how they work, where they drift, and how to stay in command. Read them in order: from what agentic tools are, to writing work an agent can run, to keeping a whole fleet of Cursor Cloud Agents busy without conflicts.

  1. 01Step 1 · What they are

    Agentic AI coding tools, without the chaos

    What makes a coding tool 'agentic', where agents drift, and how well-scoped stories in Linear keep a whole fleet of them on target.

    Read the guide →
  2. 02Step 2 · Writing the work

    A good story isn't a long one

    What makes work a coding agent can run well: carrying your vision, closing the gaps it would otherwise guess at, and defining done before anyone starts.

    Read the guide →
  3. 03Step 3 · Fast feedback

    Give your agents instant feedback — before they push

    Agent hooks, pre-commit gates and failure messages written for the agent: catch mistakes at the edit, not twenty minutes after the push.

    Read the guide →
  4. 04Step 4 · Running many at once

    Get out of the bottleneck. Keep the craft.

    How to run several background coding agents in parallel without conflicts or wasted tokens — and stay the craftsman instead of turning into a QA robot.

    Read the guide →
  5. 05Step 5 · Reviewing the output

    Review the intent, let the machines check the rest

    When agents open more PRs than you can read, let machines check everything they can — and spend your review on intent, naming and direction.

    Read the guide →
  6. 06Step 6 · Changes across repos

    One change, many repos, no broken main

    Plan the whole change once, slice it per repo, and merge in an order where every repository stays deployable at every step.

    Read the guide →
  7. 07Step 7 · Cursor in practice

    Cursor background agents: run more of them in parallel

    What Cursor background agents are, how to start them, and what it takes to keep a fleet of them busy — with Linear as your control tower.

    Read the guide →
  8. 08Step 8 · Choosing a tool

    Cloud coding agents compared: Cursor, Claude Code, Codex and Copilot

    Where each cloud agent runs, how you start it, what it hands back and how it bills — with every claim linked to the vendor's own docs.

    Read the guide →
  9. 09Step 9 · Keeping costs down

    What a coding agent run really costs (and how to spend less)

    The price per token is the smallest lever. Runs per merged PR, task size, feedback speed and the hidden CI and review bills matter more.

    Read the guide →