idea-to-ship plugin
Six linked skills take a project from a raw idea to a validated build plan and a first draft in Claude Code and Codex.
- 91 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- autopilot and build-loop run builds, tests and a browser on their own, and audit-and-fix creates commits in the repository.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/idea-to-shipDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- autopilot and build-loop run builds, tests and a browser on their own, and audit-and-fix creates commits in the repository.
This is third-party code. Review the repository files before installing.
What it does
idea-to-ship is six skills that work together or alone. The manual tier: ideate turns an idea into a locked concept in CONCEPT_BRIEF.md and demands a success metric and a kill criterion. deep-dive runs parallel specialists, then a red team, and returns a verdict with a 1 to 10 confidence score. prompt-pack cuts the work into self-contained prompts, each of which verifies itself and leaves the app working. The autonomous tier: build-loop renders the app, takes a screenshot, critiques and rebuilds until acceptance criteria pass or a stop condition fires. autopilot runs the whole chain by itself and returns a first draft plus an honest list of what a human must finish. audit-and-fix works on an existing repository: audit, triage, fixes with a regression test and one commit per verified unit.
Who it is for. Solo builders and small teams who start coding before deciding what is worth building, working in Claude Code or Codex.
Good fit when
- You have a raw product idea and want to know if it is worth building before any code exists.
- A big task does not fit one chat and must split into independent steps.
- The built app compiles but half the screens are unwired, and you want a screenshot-based verification loop.
- An inherited repository needs an audit and fixes without breaking what works.
Not a fit when
- The task is small: a one-function edit does not justify a multi-step pipeline.
- Token limits are tight, while deep-dive fans out 4 to 6 specialists and spends tens of thousands of tokens.
- You need market validation: a synthetic persona's reaction does not count as validation, and autopilot does not stand in for it.
Example request
I have an idea for a studio service: help me decide whether to build it and lock the concept.Limitations
The author calls the autonomous tier experimental: output hits a ceiling of about 80 percent, and the last stretch of correctness, security and taste stays with a human. deep-dive spends many tokens, so run it deliberately. audit-and-fix stops at local commits and publishes nothing itself. The suite works best in Claude Code; in Codex some checks run serially and confidence in the conclusions is lower. audit-and-fix has not been tested in Codex. The last commit in the repository is dated 4 August 2026.
How to disable. Remove the plugin with /plugin or delete the copied skill folders from ~/.claude/skills and ~/.agents/skills. Delete CONCEPT_BRIEF.md files and research/ directories in projects by hand.
FAQ
How do I install idea-to-ship in Claude Code?
Run `/plugin marketplace add nelsonwerd/idea-to-ship-skills` and `/plugin install idea-to-ship@nelsonwerd`. The skills get names like `/idea-to-ship:ideate`; if they do not appear, run `/reload-plugins`.
How do I use the skills in Codex?
Copy the folders from `skills/` into `~/.agents/skills/` and restart Codex. Then describe the task in words or find a skill through `/skills`.
Can I take just one skill?
Yes. Each skill runs alone. ideate is light and fits a Claude chat; the others need repository access.
What does audit-and-fix do?
It audits an existing repository, picks what is worth fixing and applies fixes. Each fix comes with a failing regression test and its own commit, and nothing is published.
How many tokens does deep-dive spend?
A lot: a full run calls 4 to 6 specialists, then synthesis, verification and a red team, over 10 agent calls. For a narrow task use 2 or 3 lanes or an ordinary single-pass review.
Discussion
No questions or reviews yet.
Related
PluginsOrchestrator skill: from an idea the agent writes requirements and a spec, splits work into tasks, builds the project, checks it
Skills for spec-driven development: brainstorming, planning, debate-style review and idea grilling before the first commit.
The /improve skill: a strong model audits a repository and writes self-contained fix plans, and you hand the execution to a cheaper model
A Claude Code plugin for large specs: requirements with proof obligations, a walking skeleton and audited sprints
hai-stack walks a software change through five steps, from judging the idea to recording results, with one skill per step.
Skill and plugin for Codex and Claude Code: the agent drives a project through a plan-to-maintain loop with human gates.
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| idea-to-ship (this card) | no | 91 | |
| Spec Kit | no | 149,633 | drives a specification and tasks for code; idea-to-ship first checks whether to build and adds autonomous build and audit loops |
| BMad Method | no | 55,338 | builds agile-team roles and documents; idea-to-ship is six standalone skills with a kill criterion and screenshot checks |
README in short
The skills cover idea, validation, plan, build and audit. The repository holds the skills, the plugin manifest, .skill archives for Claude chat and a write-up of a 13.5-hour autonomous run.