Skip to content
Back to search

epatel

MarketplaceWorks with Claude CodeGitHub

epatel's Claude Code plugins

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/epatel--project-setup/badge.svg)](/listings/epatel--project-setup)

HTML

<a href="/listings/epatel--project-setup"><img src="/api/listings/epatel--project-setup/badge.svg" alt="findsafeskills safety rating" /></a>

Install

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

Do these commands in a shell:

claude plugin marketplace add epatel/project-setup
claude plugin install project-setup@epatel --scope user

The option `--scope user` makes the skill available in all your projects. If you do not use this option, the skill is available only in the current project.

- **To examine the installation:** Do `claude plugin list`. Make sure that the list shows `project-setup@epatel` with `Scope: user`. - **To update the plugin:** Do `claude plugin update project-setup@epatel`. - **To install in a Claude Code session:** Do `/plugin install project-setup --marketplace epatel/project-setup`. The first time, this command can add only the marketplace. Do the command again until it shows "Installed". This command installs the skill only for the current project.

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

.claude-plugin/marketplace.json · Manifest · 14 lines
  1. {
  2. "name": "epatel",
  3. "owner": {
  4. "name": "epatel"
  5. },
  6. "description": "epatel's Claude Code plugins",
  7. "plugins": [
  8. {
  9. "name": "project-setup",
  10. "source": "./",
  11. "description": "AI-runnable project setup from setup-todo-list.md (run, doctor, author)."
  12. }
  13. ]
  14. }
README.md · README · 128 lines
  1. # AI-runnable project setup
  2. [![Audit](https://github.com/epatel/project-setup/actions/workflows/audit.yml/badge.svg)](https://github.com/epatel/project-setup/actions/workflows/audit.yml)
  3. This repository gives a format and a Claude Code skill for project setup instructions. An AI agent can do the setup steps, continue a setup that stopped, and examine the setup again later. The last function is similar to `flutter doctor`.
  4. ```mermaid
  5. flowchart LR
  6. M["setup-todo-list.md<br/>(committed, maintainers)"] -- "agent does steps" --> L["setup-todo-list.local.md<br/>(gitignored, one line per step)"]
  7. M -- "git pull gives new or changed steps" --> D{doctor}
  8. L --> D
  9. D --> R["[✓] ok [✗] not done or broken<br/>[!] new or changed [-] optional"]
  10. ```
  11. ## Add the setup format to a project
  12. Give this instruction to a coding agent (Claude Code, Codex, Cursor, or a different agent):
  13. > See https://raw.githubusercontent.com/epatel/project-setup/main/ADOPT.md and add project setup instructions to this repo.
  14. The agent does these tasks:
  15. 1. It copies the skill into `.claude/skills/project-setup/`.
  16. 2. It reads the repository to find all the setup steps. This includes steps that the README does not show.
  17. 3. It writes `setup-todo-list.md`.
  18. 4. It adds the local progress file to `.gitignore`.
  19. 5. It updates the README.
  20. The agent does not make a commit. You must examine the result.
  21. ## Install the skill for all your projects (Claude Code plugin)
  22. Do these commands in a shell:
  23. ```bash
  24. claude plugin marketplace add epatel/project-setup
  25. claude plugin install project-setup@epatel --scope user
  26. ```
  27. The option `--scope user` makes the skill available in all your projects. If you do not use this option, the skill is available only in the current project.
  28. - **To examine the installation:** Do `claude plugin list`. Make sure that the list shows `project-setup@epatel` with `Scope: user`.
  29. - **To update the plugin:** Do `claude plugin update project-setup@epatel`.
  30. - **To install in a Claude Code session:** Do `/plugin install project-setup --marketplace epatel/project-setup`. The first time, this command can add only the marketplace. Do the command again until it shows "Installed". This command installs the skill only for the current project.
  31. ## How the format works
  32. - **`setup-todo-list.md`** is in the repository. It contains the setup steps. Each step has these parts:
  33. - an `[id]` that does not change
  34. - a `rev` number
  35. - a `gate`: `auto`, `approve` or `input`
  36. - a `required` flag
  37. - a `check` command that does not change files
  38. - **`setup-todo-list.local.md`** is not in the repository. It contains one progress line for each step, for example `- flutter-sdk: done @rev2 2026-11-21`. The agent writes a line after each step. Thus, the agent can continue a setup that stopped.
  39. - **Doctor** compares the progress lines with the list. It uses the `id` and the `rev` of each step. Thus, it finds new steps, changed steps and removed steps. Then it does the `check` commands again. Thus, it finds steps that were done but are now broken. For example, a tool was removed from the computer.
  40. - **The progress file keeps the commit hash of the list.** The command `git diff <hash> -- setup-todo-list.md` shows the changes in the list since the last setup.
  41. - **Secrets** go only into `.env` or a similar file. Do not put secrets into the list or into the progress file.
  42. - **The list starts with instructions for agents.** Thus, all coding agents can use the list. The skill makes the work more reliable.
  43. The full specification is in [`skills/project-setup/references/format.md`](skills/project-setup/references/format.md). For examples, see [`examples/setup-todo-list.md`](examples/setup-todo-list.md) and its progress file [`examples/setup-todo-list.local.md`](examples/setup-todo-list.local.md).
  44. ## Use the skill in Claude Code
  45. | Command or request | Result |
  46. |---|---|
  47. | `/project-setup` or "set up this project" | The agent makes the progress file, shows the plan, and does the steps. It obeys the gate of each step. |
  48. | `/project-setup doctor` or "check my setup" | The agent shows a report. The report shows new steps, changed steps and broken steps. |
  49. | `/project-setup author` or "make a setup list for this repo" | The agent writes `setup-todo-list.md`. It uses the README, the CI configuration, `.env.example` and other files. |
  50. With the plugin, the full name of the skill is `/project-setup:project-setup`. The short name `/project-setup` also operates. But if a repository has its own copy of the skill, the short name starts that copy.
  51. ## When to use doctor
  52. Use doctor only when necessary. Setup steps do not change frequently. The skill tells the agent not to start doctor without a request.
  53. - **Usually:** Use doctor only when you want a report.
  54. - **Fast stop:** The progress file keeps `master-commit`. This is the last commit hash of the list. If the hash is the same and the list has no changes that are not committed, the agent stops.
  55. - **After a pull:** You can add this hook. It shows a message only when the list changed.
  56. ```bash
  57. # .githooks/post-merge
  58. git diff --quiet ORIG_HEAD HEAD -- setup-todo-list.md ||
  59. echo "setup-todo-list.md changed — consider: setup doctor"
  60. ```
  61. **WARNING:** The `check` commands come from the repository. They run in your shell. If you do not trust a repository, read its `check` commands before an agent runs doctor.
  62. ## Rules for maintainers
  63. The rules are in [the maintainer rules in format.md](skills/project-setup/references/format.md#maintainer-rules). In summary:
  64. - Do not use an `id` again.
  65. - Increase `rev` only when the meaning of a step changes.
  66. - Make sure that each `check` command is fast and does not change files.
  67. - Use `auto` only for safe steps in the project.
  68. ## Repository layout
  69. ```
  70. .claude-plugin/ plugin.json and marketplace.json (this repository is both)
  71. skills/project-setup/ the skill (SKILL.md, references/format.md)
  72. ADOPT.md instructions for an agent that adds the format to a project
  73. examples/ example setup list and progress file
  74. .github/ audit workflow and audit script (only for CI)
  75. ```
  76. ## Audit
  77. The audit workflow runs for each push, for each pull request, and one time each week. It does these checks:
  78. - The example lists obey format v1.
  79. - Each `check` command has correct shell syntax (`bash -n`).
  80. - The agent instructions in each example list are the same as the instructions in format.md.
  81. - The plugin, the marketplace and the skill are correct (`claude plugin validate --strict`).
  82. - The relative links are correct.
  83. - The download URLs in ADOPT.md are available.
  84. ## Releases
  85. ADOPT.md downloads the skill from a release tag. The current tag is **`v1.1.1`**. Thus, projects do not get changes that are not released. Tags do not move.
  86. To make a release:
  87. 1. Change `version` in `.claude-plugin/plugin.json`.
  88. 2. Change the tag in the two URLs in ADOPT.md.
  89. 3. Commit the changes.
  90. 4. Add the tag `vX.Y.Z`.
  91. 5. Push the branch and the tag together: `git push origin main vX.Y.Z`.

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.