# AI use cases for Tasktrooper (makifbaysal/tasktrooper)

Repo: [makifbaysal/tasktrooper](https://github.com/makifbaysal/tasktrooper) | Use cases: 4 | Page: https://dedupedai.com/use-cases/makifbaysal-tasktrooper/

TaskTrooper is a local-first desktop agent platform where a Kanban board hands each task to a role agent that runs an agent CLI or model API on your own machine.

This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL.

- Turn a feature request into a reviewed pull request: https://dedupedai.com/use-cases/makifbaysal-tasktrooper/board-task-to-pull-request/
- Keep a Claude Code build running through usage limits: https://dedupedai.com/use-cases/makifbaysal-tasktrooper/overnight-build-survives-usage-limit/
- Split one board across Claude Code, Cursor and OpenCode: https://dedupedai.com/use-cases/makifbaysal-tasktrooper/mixed-cli-team-on-one-board/
- Run board agents on a local model with no cloud account: https://dedupedai.com/use-cases/makifbaysal-tasktrooper/local-model-agents-no-cloud/

## How to turn a feature request into a reviewed pull request with Tasktrooper

Repo: [makifbaysal/tasktrooper](https://github.com/makifbaysal/tasktrooper) | Difficulty: Beginner | Time: 1 hour | Basis: From the README

Put a task on the board and let role agents clone, build, test and open a PR.

**Who it is for:** A solo developer with a backlog of small features who wants the work done while they do other things.

**The problem:** Writing, testing and opening a pull request for every small feature takes time that a solo developer would rather spend on decisions.

**The outcome:** A task moves through the board's columns on its own and ends as a pull request on GitHub that you can review.

### Steps

1. Install on macOS with `curl -fsSL https://raw.githubusercontent.com/makifbaysal/tasktrooper/main/scripts/install.sh | bash`, or on Windows run the `TaskTrooper- -setup.exe` from Releases.
2. Finish the guided first run: connect an agent CLI or an API provider key, then connect GitHub and import your project.
3. Create a task in Backlog with acceptance criteria describing the feature.
4. Leave the task on the board; the agent for each column picks it up and hands it to the next column.
5. When the task reaches code review, open the pull request on GitHub and review it.

### Prompt to paste into your agent

```text
Using the TaskTrooper repo at makifbaysal/tasktrooper, set up the desktop app on this machine and import the project at <project path>. Create one Backlog task titled <feature title> with the acceptance criteria <criteria>. Report the task's column and the pull request URL once it appears, and check that the task has a branch of its own.
```

### Caveat

The app is ad-hoc signed and not notarized yet, and the Windows and Linux builds are not code-signed, so expect extra confirmation steps on first launch.

Risk, remote script: The install command pipes a script from GitHub into bash, so read it first or download the .dmg from Releases instead.

Risk, credentials: The setup asks for GitHub access and an agent or model API key, so use tokens scoped to the single project you import.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/makifbaysal-tasktrooper/board-task-to-pull-request/

## How to keep a Claude Code build running through usage limits with Tasktrooper

Repo: [makifbaysal/tasktrooper](https://github.com/makifbaysal/tasktrooper) | Difficulty: Intermediate | Time: 1 hour 30 minutes | Basis: From the README

Queue work that resumes automatically after a Claude Code usage limit resets.

**Who it is for:** A developer who runs Claude Code on large tasks and loses progress when the usage limit cuts a session short.

**The problem:** When an agent CLI runs out of its usage limit mid-task, the work is often lost and must be restarted by hand.

**The outcome:** The interrupted task waits on Blocked, then resumes on its own with the same session once the limit resets.

### Steps

1. Install TaskTrooper and finish the first run with Claude Code connected as the agent CLI.
2. Import your project and put several larger tasks in Backlog or the column that your developer agent works.
3. Leave the board running; if the limit is hit, the task moves to Blocked instead of failing.
4. Check the board later and confirm the blocked task has resumed and kept its session.

### Prompt to paste into your agent

```text
In the makifbaysal/tasktrooper repo, confirm that Claude Code is connected as the agent CLI on this machine. Put the task <task title> on the board and report whether it moves to Blocked or continues after a usage limit, and whether the session resumes with --resume.
```

### Caveat

Resume behavior is described for Claude Code only, so other CLIs continue after a reset but may not keep the earlier session.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/makifbaysal-tasktrooper/overnight-build-survives-usage-limit/

## How to split one board across Claude Code, Cursor and OpenCode with Tasktrooper

Repo: [makifbaysal/tasktrooper](https://github.com/makifbaysal/tasktrooper) | Difficulty: Intermediate | Time: 1 hour 15 minutes | Basis: From the README

Assign each role agent to the CLI that suits it on a single board.

**Who it is for:** A developer who has more than one agent CLI installed and wants a developer, reviewer and QA agent to use different tools.

**The problem:** Forcing every role onto one agent CLI leaves some work done by a tool that is not the best fit for it.

**The outcome:** A board where the developer runs on Claude Code, the reviewer on the Cursor CLI and QA on OpenCode, each in its own session.

### Steps

1. Install TaskTrooper and connect Claude Code, the Cursor CLI and OpenCode during first run.
2. Open the agent settings and set the runtime for backend-developer to Claude Code.
3. Set the runtime for system-architect or security-agent to the Cursor CLI.
4. Set the runtime for qa-agent to OpenCode, then move a test task into the QA column to confirm the handoff.

### Prompt to paste into your agent

```text
In the makifbaysal/tasktrooper repo, open the agent settings and assign the runtime for backend-developer, security-agent and qa-agent to <runtime for each>. Report which runtime each agent used on its last run and confirm that each run has its own session.
```

### Caveat

The runtime is set per agent, so a role runs on one CLI at a time and you must reassign it manually to change tools.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/makifbaysal-tasktrooper/mixed-cli-team-on-one-board/

## How to run board agents on a local model with no cloud account with Tasktrooper

Repo: [makifbaysal/tasktrooper](https://github.com/makifbaysal/tasktrooper) | Difficulty: Advanced | Time: 1 hour 30 minutes | Basis: From the README

Point agents at Ollama or LM Studio so code stays on your machine.

**Who it is for:** A developer who cannot send source code to a hosted model and needs agents that run against a local model server.

**The problem:** Most agent setups call a hosted API, which may not be allowed for private or regulated code.

**The outcome:** Agents answer from a local OpenAI-compatible endpoint, and tasks progress on the board without a cloud model.

### Steps

1. Start your local model server, such as Ollama or LM Studio, and note its OpenAI-compatible endpoint address.
2. Install TaskTrooper and choose any OpenAI-compatible endpoint as the API provider during first run.
3. Assign that provider to one or more agents in the agent settings.
4. Put a small task on the board and confirm it completes against the local endpoint.

### Prompt to paste into your agent

```text
In the makifbaysal/tasktrooper repo, set an OpenAI-compatible provider pointing at <local endpoint URL> for the product-manager agent. Create a small Backlog task and report whether the agent's run completed and which model answered.
```

### Caveat

Local models must be served separately, and the README does not say which local model works best with the role agents.

Risk, credentials: If the local endpoint needs a key, store it only in the TaskTrooper provider settings and not in the repo.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/makifbaysal-tasktrooper/local-model-agents-no-cloud/