Break one goal into a sequenced coding plan

Ordewell turns a single objective into a step-by-step task list for coding agents, with dependencies, review, and parallel runs.

Done with Custom agentยท Intermediate

What they did
They started by giving Ordewell one target to work toward. A planner agent, running without write access, examined the repository and asked clarifying questions, then produced a plan that assigned each step its own runner, model, effort setting, and operating mode. The plan could be edited before launch; after approval, Ordewell opened separate coding-agent runs for the AI steps, used concurrency where the dependency chain allowed, and carried prior outputs into later work.
How it went
In the README example, a rate-limiting objective was split into four steps: three handled by AI and one done manually. The job ended with 4 completed and 0 failed. A step only registered as finished when its completion marker appeared in the runner output.
Tips
The TUI relies on tmux, so Windows users need to run it through WSL. By default it keeps up to 3 sessions going at once, and that limit can be adjusted between 1 and 5.

Try it yourself with Claude Code

I have one objective: [describe your goal]. Look through [repo path] in read-only mode, then draft a numbered plan that breaks the work into [number] concrete tasks. For each task, include its dependencies, which runner should handle it, what model and effort level to use, and how we can tell it is complete. Save the plan to [PLAN file] and stop before changing any files. If a step would involve sending, deleting, or submitting anything, pause and ask me to approve that action first.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Read the original on github.com โ†—

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.