Apache 2.0 · Eight supported coding agents

Stop juggling agent tabs. Talk to one planner.

It runs Claude Code, Codex and six more agents for you, in plain sight, and nothing runs until you approve.

macOS and Linux · one binary
curl -fsSL https://ordewell.ai/install.sh | sh

Describe what you want built. Ordewell reads your repo, asks about anything left vague, and returns a plan of tasks you can edit. Build tasks wait for your approval. Each change works on its own branch, and you read the diff before you merge.

New in 0.13: install without Node.js →

No Node.js needed on macOS and Linux · Windows installs with npm · no API key if you already have a coding agent

A goal fans out into six tasks joined by dashed dependency edges, running top to bottom. Each task names its own runner (claude, codex or opencode), and every task closes with DONE. Completion is reported by the runner itself, and the dependency graph decides what runs in parallel. goal audit auth flow claude DONE ✓ refactor sessions codex DONE ✓ update README claude DONE ✓ harden middleware codex DONE ✓ fix flaky tests opencode DONE ✓ integration pass claude DONE ✓
01

Say what you want

Ordewell reads your repo and asks about anything you left open.

02

Shape the plan

Change any task, agent or model before anything runs.

03

Read the diff, then merge

Change tasks work side by side on their own branches. Yours waits for you.

How it works

You see the plan before any build task starts.

Every task is written out with the agent, model and effort that will run it. You can change any of it first.

01 · Plan

It reads, then asks

The planner reads your repo without changing it, asks about anything unclear, and writes the plan: ordered tasks, each naming the agent and model that will run it.

02 · Run

Many agents, one plan

Each change task gets its own branch and worktree. Independent tasks run at the same time. A dependent task waits until its inputs are on the run's integration branch.

03 · Verify

Done means reported

A task is done when its agent reports completion through Ordewell's own tools, or when a script's exit status says so. No second model judges the work. A task that never reports fails where you can see it.

Why it's different

Four things Ordewell does on purpose.

Approval

Build tasks wait for you

The plan is yours to rewrite before anything builds: prompts, agent, model, effort, mode and dependencies.

Evidence

Done means reported

No second model is asked whether the work looks done. The agent's own completion report decides, and a script's exit status decides for scripts.

Mixed

Several agents, one plan

Eight supported agents can share a plan. Each task picks its agent, model, effort and mode, and a task can move to another agent.

Limits

Limits pause work, they don't end it

A task that hits its agent's usage limit pauses. Claude Code and Codex name the reset, and the task continues in its saved session then, while Ordewell is running. After a restart it waits for you. Move a paused task to another agent and its unfinished work comes with it.

VS Code extension

The whole loop, beside your code.

Live thinking, every research step, and the plan itself, pinned above the composer.

Install

bash
code --install-extension ordewell.ordewell

Or search Ordewell in the Extensions view, or open the Marketplace listing. In Cursor, Windsurf or VSCodium, get it from Open VSX. The extension bundles its own core, so there is nothing to install from npm.

First run

  1. Open a project folder.
  2. Open the Ordewell panel from the activity bar.
  3. Type a goal. Build tasks wait until you approve the plan.

Choose an installed coding agent to plan for you, read only, using its own credentials.

Captured from Ordewell 0.11.0. Model names and interface details may have changed since.

The Ordewell panel in VS Code. The goal Add rate limiting to the public API, the research folded into one line reading Listed a directory, read 2 files, searched the code, the planner's question whether limits apply per API key or per client IP with its recommendation, the answer Per key with the IP fallback, and a Plan generated, 5 tasks marker.
The conversation Your goal, the research folded into one line you can open, and the question the planner asked before it planned.
The plan docked above the composer: an Execute Plan button over five task cards, each with its runner and model, and an effort level where the model offers one, such as claude-code, Claude Sonnet 4.5, high and opencode, DeepSeek V4 Flash, a manual rollout task, and dependency badges. Under the plan, the footer reads Planner ctx 38k (19%), Claude 5h 23% used, Weekly 41% used.
The plan Docked above the composer. Each agent task names its runner and model before you press Execute Plan. The footer shows the planner's context and your account's limits.
An open task card's assignment: Runner set to OpenCode, Model DeepSeek V4 Flash on openrouter, and Mode Build: Edit files and run commands.

Move a task to another agent

Switch a task's runner and its model, effort and mode follow, picked from what that agent actually offers. An assignment that couldn't start never gets saved.

The same task's model picker open on OpenCode's catalog: a search field above DeepSeek V4 Flash, Qwen3 Coder Next, GLM-5 and Kimi K3.

Only models that runner can start

The picker lists the selected runner's own catalog, so you never type an id that only fails once the task starts.

A task's model field reading Claude Sonnet 4.5 on anthropic, beside a thinking effort select set to high.

Thinking effort is per task too

The levels offered are the ones that model actually has. A docs task doesn't need to think as hard as a migration, and you decide which is which.

A task card's skills: tdd and api-conventions attached as removable chips, with an Add skill menu under them.

Skills your whole team shares

Write down how your team works once, as a skill in your repo. The planner attaches it to the tasks it fits, and you see it on every task card. /grilling, /to-spec and /tdd come built in.

Watch a real run in VS Code, 42 seconds: a goal becomes a plan, three tasks run at once, and the supervisor reports back. Sped up where marked. Recorded on October 20, 2026 with Ordewell 0.10.10.

Terminal UI

The same loop in your terminal, over SSH.

Conversation on the left, live plan on the right. Single keys drive the plan: no menus, no mouse.

Install

bash
curl -fsSL https://ordewell.ai/install.sh | sh
ordewell   # in your project

One binary, no Node.js. brew install ordewell/tap/ordewell works too. On Windows, or if you prefer npm, run npm install -g ordewell (Node.js 20.3 or newer; mouse and paste on Windows need 22.17).

First run

  1. Open a project directory in a terminal.
  2. Run ordewell. The opening screen names the planner and the runners it found.
  3. Type a goal. Build tasks wait until you approve the plan.

An installed coding agent plans by default, with no API key. Change it in place with /planner and /runners.

Ordewell's terminal UI: the conversation on the left lists the planner's reads and a question about reusing a shared Redis client, and the plan on the right lists seven tasks, six with their runner and model and their effort where the model offers one, and one manual rollout task; two are done and one is running. The footer reads Planner ctx 38k (19%), Claude 5h 23% used, Weekly 41% used, and the status bar reads Guarded · supervisor Watch.
The planner conversation and the live plan, side by side. Guarded is the default autonomy level.
  • tab swap chat and plan
  • f start selected task
  • E run the whole plan
  • R change runner
  • o change model
  • M execution mode

CLI

For a pipeline instead of a person.

bash
# plan on an agent you already have, no API key
export AI_PROVIDER="claude-code"        # or codex, opencode and five more

ordewell plan --goal "Add rate limiting to the public API"
ordewell task-model 3 sonnet            # change a task before it runs
ordewell run
ordewell handoff review                 # read the diff first
ordewell handoff merge                  # then bring it onto your branch

Get started

One install, one goal.

Terminal

One binary

bash
curl -fsSL https://ordewell.ai/install.sh | sh

Installs to ~/.local/bin on macOS and Linux, with no Node.js. Or brew install ordewell/tap/ordewell. Then run ordewell in your project. On Windows, or if you prefer npm, use npm install -g ordewell (Node.js 20.3 or newer); npx ordewell runs it without installing.

VS Code

The extension

bash
code --install-extension ordewell.ordewell

Or search Ordewell in the Extensions view. It bundles its own core, so there is nothing to install from npm.

Get it on the Marketplace →

What you need

  • One coding agent, installed and signed in: Claude Code, Codex or one of six more. It runs tasks and can plan too.
  • No Node.js on macOS and Linux: the terminal UI and CLI are one binary. Windows installs with npm and needs Node.js 22.17 or newer for the terminal UI's mouse and paste, plus Git for Windows. The VS Code extension needs only VS Code.
  • git is optional. Without git, or outside a git repository, every task works directly in the folder. git 2.38 or newer gives each task its own worktree, delegated tasks and Merge all.

Ordewell is pre-1.0, so minor releases can include breaking changes. Read the changelog before you upgrade.

Read the docs →

Read the plan before you pay for it.

Open source, Apache 2.0, and the plan is yours to rewrite.