Task-centric workflow
Create a task against a repository, choose a provider, model, and permission mode, and keep the prompt, run history, diff, review, preview, and GitHub status attached to that task.
Task Monki
Task Monki is a macOS desktop app for managing coding-agent tasks from request to pull request, with local previews, reviews, and draft pull request publishing.
Task Monki at a glance
Task Monki is a desktop app for managing coding-agent work from request to pull request. It is designed to keep each task’s agent run, code changes, preview, review, and pull request together so you can follow the work in one place.
The product is aimed at local development workflows. It runs on macOS and works with supported agent runtimes such as Codex and OpenCode, with ACP compatibility for Cursor Agent, Grok Build, and Claude Agent. The app also supports local previews when a recipe is available, letting you inspect and compare results before publishing a branch or draft pull request.
Core capabilities
Dedicated worktrees and branches
Task Monki prepares a dedicated worktree and Git branch for each task so multiple agents can work on the same repository without sharing one directory.
Agent activity tracking
Follow plans, tool activity, approvals, logs, messages, changed files, and reported test output while the agent works.
Local preview support
Open an approved local preview for supported tasks, with declared services, readiness checks, and stable local routes; replace the preview after source changes.
Independent review and fixes
Request a separate read-only review, read the findings in context, and send actionable issues back to the implementation agent for fixes.
Publish to GitHub
Commit or publish the branch, open a draft pull request, and follow its current GitHub checks and reviews. Task Monki itself does not merge pull requests.
Practical ways to use Task Monki
Run one coding task from start to finish
Create a task for a repository, assign an agent, and keep the resulting worktree, diff, and run history tied to the same task while the agent works.
Coordinate parallel agent work
Use dedicated branches and worktrees to let several agent sessions work on the same repository without sharing a single directory.
Inspect changes in a local preview
Open a local preview for a supported task, check the resulting behavior, and replace the preview after source changes without rebuilding the setup by hand.
Use a review-and-fix loop
Ask a second agent to review the implementation, compare the findings with the code and test output, and send actionable issues back for fixes.
Prepare delivery to GitHub
Commit or publish the task branch and open a draft pull request once the result is ready to share on GitHub.
Pros and Cons
Pros
- Keeps the task, branch, diff, review, preview, and GitHub status attached in one place.
- Supports multiple agent runtimes, including direct support for Codex and OpenCode plus ACP compatibility for other agents.
- Creates dedicated worktrees and branches, which helps separate concurrent tasks on the same repository.
- Can open local previews with declared services and readiness checks when a supported recipe is present.
- Supports a review-and-fix loop where findings can be sent back to the implementation agent.
Cons
- The alpha is not ready for critical repositories, so the docs recommend using repos you can recover and reviewing every change before shipping.
- Preview startup and replacement currently require Task Monki on macOS.
- Task Monki does not merge pull requests, so delivery still requires a human decision at the end.
FAQ
What do I need to run Task Monki?
Task Monki runs locally on macOS. For previews, it also requires Docker when a recipe uses managed PostgreSQL, Redis, or Docker Compose.
Which agent runtimes and providers are supported?
Task Monki supports Codex and OpenCode directly, and it can connect to Cursor Agent, Grok Build, and Claude Agent through ACP compatibility integrations.
Can one task go all the way from request to pull request?
Yes. The documentation says Task Monki can prepare a dedicated worktree, open a local preview when supported, request an independent review, and then publish a draft pull request.
Does Task Monki merge pull requests automatically?
No. Task Monki does not merge pull requests. You can commit, publish, and open a draft pull request, but the final decision still stays with you.
Is the alpha safe for critical repositories?
No. The alpha is not ready for critical repositories, and the docs recommend using repositories you can recover and reviewing every change before shipping it.
Quick Facts
- Category
- Developer Tool
- Platform
- macOS desktop app
- License
- MIT
- Supported runtimes
- Codex, OpenCode, and ACP-compatible Cursor Agent, Grok Build, Claude Agent
- Workflow
- Request to pull request with review and local preview support
- Source domain
- monki.work
Task Monki Alternatives
SonOf
SonOf connects to your repo and PM tool, audits your codebase, and turns approved work into shipped tickets with senior engineering review.
ByteAsk
ByteAsk is a terminal-first AI coding agent for C and C++ that edits repos and verifies changes with compilers, debuggers, sanitizers, and tests.
CreateOS Sandbox
CreateOS Sandbox is an isolated compute environment for running code and agent workloads in Firecracker micro-VMs with private networking and SDK, CLI, or MCP control.
hob
hob is an independent workspace for coding agents, with local control over sessions, terminals, history, routing, and follow-up work.
Manta AI
Manta AI is an autonomous web app testing tool that maps app behavior, catches regressions, and generates tests from a URL, no scripts or selectors needed.
Ghost
Ghost is a terminal-based AI assistant for chatting, code generation, and CLI tasks. Includes free models, supports Linux, macOS, Windows, and is open source.