Plan and verify parallel coding-agent tasks with Ordewell
Use Ordewell when an operator wants one goal turned into an editable dependency graph of coding-agent tasks, each with its own runner, model, branch, completion marker, and review handoff.
npx skills add agentskillexchange/skills --skill plan-and-verify-parallel-coding-agent-tasks-with-ordewell
Ordewell is task orchestration for coding agents across Claude Code, Codex, OpenCode, and provider-backed planners. Use it when an operator wants a planner to read a repository without writing to it, ask clarifying questions, produce an editable dependency graph, assign each task to a runner and model, execute independent work in isolated git worktrees, and verify task completion from runner output markers before anything lands. Invoke this instead of using a coding agent normally when the useful workflow is repeatable plan-run-verify-handoff control: run ordewell plan with a goal, inspect or edit task prompts, runners, models, modes, and dependencies, run the plan, review the handoff diff, then merge or discard the integration branch. The scope boundary is editable coding-agent task planning, isolated execution, marker-based verification, and human handoff, not a generic multi-agent framework, terminal UI, VS Code extension, or package listing.
What this skill actually does
Inputs and prerequisites: Ordewell, Node.js 20 or newer, git, at least one supported runner such as Claude Code, Codex, or OpenCode, and tmux for the terminal UI.
Setup notes: Install the CLI with npm install -g ordewell, or install the Ordewell VS Code extension. Then run ordewell in a project, select planner and runners with /planner and /runners, or use the CLI sequence ordewell plan –goal “…”, ordewell run, ordewell handoff review, and ordewell handoff merge.
Source and verification boundary: use https://ordewell.ai/docs as the canonical reference before running the workflow; keep commands, API calls, CLI usage, and generated outputs reviewable against that upstream source.
Framework fit: publish this as a Multi-Framework workflow only when the operator can invoke the documented toolchain directly, rather than treating the upstream project as a generic product listing.