Turn one goal into an ordered plan of coding-agent tasks
Stop juggling agent tabs. Task: Stop juggling agent tabs. Talk to one planner. It runs Claude Code, Codex and OpenCode for you, in plain sight. Review build plans before they run, or turn on autorun. Website · Documentation · News Describe what you want to build. Ordewell explores your repository, asks clarifying questions, and turns your goal into a plan you can read and edit. Run independent tasks in parallel, review the changes, and merge into your branch. Submitted build plans wait for your approval by default; run settings let you automate execution and merging. Claude Code, Codex and OpenCode are supported out of the box, and can be mixed freely within one plan. Try it in your project, with nothing to install: This README describes 0.12.3. See the published releases for versions available to install and their release notes. Ordewell is pre-1.0, so minor releases can include breaking changes. Read the changelog before you upgrade. Features Installation npx ordewell runs it without installing. To keep it: For VS Code, install Ordewell from the Marketplace or run code --install-extension ordewell.ordewell. In Cursor, Windsurf or VSCodium, get it from Open VSX. The extension bundles its own core and needs nothing from npm. Requirements: Node.js 20.3 or newer, at least one of Claude Code, Codex or OpenCode, and git 2.38 or newer for task isolation. Windows The VS Code extension runs natively on Windows. The terminal UI has not been verified natively; run it under WSL. See the platform notes for setup and current limitations. Quick start Run ordewell in your project to open the terminal UI, then type a goal. The first run lets you pick a planner and runners with /planner and /runners. In VS Code, open your project folder and the Ordewell panel, then type a goal. See the extension guide for setup. Ordewell also supports command-line workflows for scripting. Tried it? Tell us how your first run went. See it in action A real run, sped up where marked. Watch the full 48 seconds. How it works Describe your goal. The planner explores your workspace without modifying it and asks clarifying questions. Review the plan. Edit tasks, dependencies, runners and models before approving the build plan. Run the tasks. Independent tasks run in parallel; dependent tasks wait for the work they need. Follow progress live. Review and merge. Read the diff and test evidence, then merge the changes into your branch or discard the run. Autorun and automatic merging are off by default. During planning, research tasks can start without build-plan approval, and script tasks can run with consent under your approval and run settings. See run settings. The planner only reads; tasks run as you, in the mode the plan assigns. See SECURITY.md for what each part can touch. Without a git repository or with isolation disabled, tasks share the workspace; some files can also be shared between worktrees. See isolation and handoff for details. Documentation The documentation covers the planner options, skills, commands, configuration and platform notes. Design decisions, including the options that were rejected, are recorded as architecture decision records, and CONTEXT.md defines the project's vocabulary. For how the planner and the runners each get their models and credentials, see planners and runners. To write your own planner or task skills, see the skills guide. If Codex's sandbox cannot start on your machine, see troubleshooting. For the experimental container mode, off in published versions, see running tasks in containers. Where it's going Next: bounded, standing consent for running on its own (#23), an escalation inbox across surfaces (#30) and a daemon that resumes plans after a restart (#31). In 0.11.0: a planner that can start consented script tasks (#107), read with a sandboxed shell that can loop and script (#64, #109), keep talking while research tasks report back (#111), and plan check-and-fix loops (#118). Shipped in 0.10: the supervisor, which watches a run and, if you let it, acts within limits you set. The full roadmap, and where to help, is in #21. Contributing Contributions are welcome. Please read CONTRIBUTING.md for the build order and project layout, and report security issues through SECURITY.md rather than the public tracker. Good places to start are the issues labelled good first issue and the runner integrations that are open for help. Acknowledgements The grilling, spec, architecture and TDD skills are adapted from Matt Pocock's skills (MIT).; tools: custom-built or unnamed agent. Human role: The original post does not say how a person stayed involved.; limitations: Requires Node.js 20.3+ (or specific installers), git 2.38+ for isolation, and at least one supported coding agent. Windows terminal UI is unverified natively; use WSL. Autorun and automatic merging are off by default to ensure human oversight. Pre-1.0 status implies potential breaking changes in minor releases..
Tools
custom-built or unnamed agent
Published on Agent Work Atlas:
Overview
Stop juggling agent tabs.
Workflow
Stop juggling agent tabs. Talk to one planner. It runs Claude Code, Codex and OpenCode for you, in plain sight. Review build plans before they run, or turn on autorun. Website · Documentation · News Describe what you want to build. Ordewell explores your repository, asks clarifying questions, and turns your goal into a plan you can read and edit. Run independent tasks in parallel, review the changes, and merge into your branch. Submitted build plans wait for your approval by default; run settings let you automate execution and merging. Claude Code, Codex and OpenCode are supported out of the box, and can be mixed freely within one plan. Try it in your project, with nothing to install: This README describes 0.12.3. See the published releases for versions available to install and their release notes. Ordewell is pre-1.0, so minor releases can include breaking changes. Read the changelog before you upgrade. Features Installation npx ordewell runs it without installing. To keep it: For VS Code, install Ordewell from the Marketplace or run code --install-extension ordewell.ordewell. In Cursor, Windsurf or VSCodium, get it from Open VSX. The extension bundles its own core and needs nothing from npm. Requirements: Node.js 20.3 or newer, at least one of Claude Code, Codex or OpenCode, and git 2.38 or newer for task isolation. Windows The VS Code extension runs natively on Windows. The terminal UI has not been verified natively; run it under WSL. See the platform notes for setup and current limitations. Quick start Run ordewell in your project to open the terminal UI, then type a goal. The first run lets you pick a planner and runners with /planner and /runners. In VS Code, open your project folder and the Ordewell panel, then type a goal. See the extension guide for setup. Ordewell also supports command-line workflows for scripting. Tried it? Tell us how your first run went. See it in action A real run, sped up where marked. Watch the full 48 seconds. How it works Describe your goal. The planner explores your workspace without modifying it and asks clarifying questions. Review the plan. Edit tasks, dependencies, runners and models before approving the build plan. Run the tasks. Independent tasks run in parallel; dependent tasks wait for the work they need. Follow progress live. Review and merge. Read the diff and test evidence, then merge the changes into your branch or discard the run. Autorun and automatic merging are off by default. During planning, research tasks can start without build-plan approval, and script tasks can run with consent under your approval and run settings. See run settings. The planner only reads; tasks run as you, in the mode the plan assigns. See SECURITY.md for what each part can touch. Without a git repository or with isolation disabled, tasks share the workspace; some files can also be shared between worktrees. See isolation and handoff for details. Documentation The documentation covers the planner options, skills, commands, configuration and platform notes. Design decisions, including the options that were rejected, are recorded as architecture decision records, and CONTEXT.md defines the project's vocabulary. For how the planner and the runners each get their models and credentials, see planners and runners. To write your own planner or task skills, see the skills guide. If Codex's sandbox cannot start on your machine, see troubleshooting. For the experimental container mode, off in published versions, see running tasks in containers. Where it's going Next: bounded, standing consent for running on its own (#23), an escalation inbox across surfaces (#30) and a daemon that resumes plans after a restart (#31). In 0.11.0: a planner that can start consented script tasks (#107), read with a sandboxed shell that can loop and script (#64, #109), keep talking while research tasks report back (#111), and plan check-and-fix loops (#118). Shipped in 0.10: the supervisor, which watches a run and, if you let it, acts within limits you set. The full roadmap, and where to help, is in #21. Contributing Contributions are welcome. Please read CONTRIBUTING.md for the build order and project layout, and report security issues through SECURITY.md rather than the public tracker. Good places to start are the issues labelled good first issue and the runner integrations that are open for help. Acknowledgements The grilling, spec, architecture and TDD skills are adapted from Matt Pocock's skills (MIT).
Reported result
No measured business outcomes or success metrics are reported in the source. The tool is pre-1.0, so stability and breaking changes are expected. A proposed acceptance check is to verify that a sample task plan is generated, approved, and executed without unauthorized file modifications.
Human role
The original post does not say how a person stayed involved.
Limitations
Requires Node.js 20.3+ (or specific installers), git 2.38+ for isolation, and at least one supported coding agent. Windows terminal UI is unverified natively; use WSL. Autorun and automatic merging are off by default to ensure human oversight. Pre-1.0 status implies potential breaking changes in minor releases.
Try it
Act as a developer using Ordewell to plan a feature. 1. Install Ordewell via `npx ordewell` or the VS Code extension. 2. Open your project [replaceable: project_path] and type a goal: [replaceable: feature_description]. 3. Review the generated plan, edit tasks if needed, and approve it. 4. Monitor execution and review the diff before merging. 5. Confirm no files were changed without approval. If inputs are missing, ask for the specific feature goal and project path. Do not enable autorun for this test.
Source reports are not independently verified. Publication date is the catalog publication date, not the original article date.