Morning Star Harness plugin
Morning Star drives agent work from request to merge through a PM agent, plan review and acceptance, with a decision log
- 69 ★ on the repository
- Last commit
- MIT
Broad access
Why this level
- Agents write code on branches and drive work to a pull request, and install adds files to the project and host settings
- Several roles and reviewers per plan spend tokens many times over
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/morning-star-harnessDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- Agents write code on branches and drive work to a pull request, and install adds files to the project and host settings
- Several roles and reviewers per plan spend tokens many times over
This is third-party code. Review the repository files before installing.
What it does
Morning Star brings a delivery process to AI tools such as dsh, omp, OpenCode, Cursor, Kimi Code, ZCode and Codex. You enter the PM role once and describe the task in your own words. A `project-manager` agent clarifies the request, keeps a plan and carries the work through requirements, implementation, independent review, acceptance and close. Separate roles own requirements, architecture, implementation, quality control, acceptance, audits and operations, each with a narrow brief. Every plan gets plan-level review, by default from three independent seats, and multi-task plans get per-task review as well. An iteration closes only after the merge. Plans, findings, reports and decisions are written to the harness directory of the repository, so the work history is visible in the repository. Audit commands read code and report without editing. `/amazing-pr-review` leaves comments on a pull request and never approves it.
Who it is for. Teams working in Cursor, Codex or OpenCode that want a plan, independent review and acceptance for every task.
Good fit when
- A feature is described in one sentence and driven to a pull request through a plan and review
- Work stops halfway and the plan keeps the place where it left off
- Independent reviewers are wanted for every plan
- A code audit or e2e check is needed on request with a report
- An iteration holds several plans and closes after the merge
Not a fit when
- You need a quick one-line fix: the process adds roles and checks
- Your working tool is Claude Code: it is not in the host list
- Plans and reports must not live in the repository
Example request
Enter the PM role and add rate limiting to the public API with tests, then carry the work to a pull requestLimitations
Install differs by host: dsh and omp need their CLIs, Cursor needs git, and Kimi Code installs the plugin by the native command only. Command names depend on the host, some carry a plugin prefix, and Codex installs command skills at project scope only. The process is heavy and spends tokens: several roles join and three reviewers check each plan. `/amazing-e2e-check` runs only on request.
How to disable. For the CLI install the removal mode is described in `INSTALL.md`, for the native one run your host plugin removal command. Delete the folder with plans in the repository by hand.
FAQ
How do I install Morning Star in Cursor?
Run `npx @mstar-harness/cli init --target cursor`. By hand: clone the repository into `~/.cursor/plugins/local/morning-star-harness` and reload the window.
Which hosts are supported?
`dsh` (DeepSeek Harness), `omp`, OpenCode, Cursor, Kimi Code, ZCode and Codex. Each has two install routes: CLI and native.
Who checks the work?
Separate roles: plan review by three independent seats, per-task review and PM acceptance as the only acceptance route.
Does Morning Star merge pull requests itself?
No, an iteration closes after the merge, and `/amazing-pr-review` only comments on a pull request and never approves it.
Where is the work log kept?
In the harness directory of your repository: plans, findings, review reports and decisions.
Discussion
No questions or reviews yet.
Related
PluginsHarness drives a task through plan, work and review: spec and Plans.md live in the repo and a separate reviewer checks the result.
Agent Harness Kit: structured agent workflow
Agent Harness Kit
The ahk CLI sets up five agent roles, a task MCP server and spec skills in your project for Claude Code, Codex and Cursor
A DiDi plugin for Codex and Claude Code that runs a task from proposal to archive and keeps verified project knowledge in the repo
Repository Harness installs an AGENTS.md entrypoint, a document map and explicitly invoked onboarding skills into a repo and updates them safely
A plugin marketplace and installer for the full dev loop: task selection, review, cleanup, PRs and releases
ADLC Harness takes a task from intent to PR with one command, and the human approves only four key gates.
README in short
The README briefly states the process idea and its strengths. A table follows with install for seven hosts in two variants, sections on a single task, an iteration and audits, the workflow diagram and scope notes.