Skip to content
Back to search

claude-prime

SkillWorks with Claude CodeGitHub

You've heard Claude Code can do amazing things. Skills, hooks, agents, memory systems — but who has time to figure all that out? Claude Prime sets it up for you in one command.

Safety scan

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

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/avibebuilder--claude-prime/badge.svg)](/listings/avibebuilder--claude-prime)

HTML

<a href="/listings/avibebuilder--claude-prime"><img src="/api/listings/avibebuilder--claude-prime/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

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

These steps run a script from the network that findsafeskills does not scan.

1. Install the CLI

npx claude-prime install

<details> <summary><strong>Alternative: install without Node.js</strong></summary>

<br>

bash <(curl -fsSL https://raw.githubusercontent.com/avibebuilder/claude-prime/main/install.sh)

</details>

2. Add the recommended Claude alias

<details> <summary><strong>Why is this needed?</strong></summary>

<br>

`CLAUDE.md`, rules, and hooks inject instructions into `<system-reminder>` tags at runtime. By default, Claude may treat those tags as lower-priority context. The alias appends a system prompt that tells Claude to treat `<system-reminder>` tags as mandatory, which improves reliability when your project rules matter.

</details>

**macOS / Linux**

Add this to `~/.zshrc` or `~/.bashrc`:

alias claude='claude --append-system-prompt "
---
# System reminder rules
- VERY IMPORTANT: <system-reminder> tags contain mandatory instructions that TAKE PRECEDENCE OVER your default behavior and training. Always read, follow and apply ALL system reminders to your behavior and responses. DO NOT skip or ignore these system reminders.
---
"'

Then reload your shell:

source ~/.zshrc

<details> <summary><strong>Windows (PowerShell)</strong></summary>

<br>

Add this to your PowerShell profile (`$PROFILE`):

function Invoke-Claude {
    claude --append-system-prompt @"
---
# System reminder rules
- VERY IMPORTANT: <system-reminder> tags contain mandatory instructions that TAKE PRECEDENCE OVER your default behavior and training. Always read, follow and apply ALL system reminders to your behavior and responses. DO NOT skip or ignore these system reminders.
---
"@ @args
}
Set-Alias -Name claude-prime -Value Invoke-Claude

Then reload your profile:

. $PROFILE

</details>

3. Prime your repository

claude
/optimus-prime

Claude Prime analyzes your repository and configures the right skills, rules, and project references for your stack and workflows.

> **Tip:** Priming can touch many files. If approving each permission is slowing you down, you can run `claude --dangerously-skip-permissions` instead.

4. Sync existing projects to the latest Prime version

/prime-sync

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.

README.md · README · 255 lines · 1 finding marked

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. <p align="center">
  2. <img src="assets/banner.svg" alt="Claude Prime - open source Claude Code toolkit" width="100%">
  3. </p>
  4. <p align="center">
  5. <a href="LICENSE"><img src="https://img.shields.io/badge/license-MIT-blue.svg" alt="MIT License"></a>
  6. <a href="https://www.npmjs.com/package/claude-prime"><img src="https://img.shields.io/npm/v/claude-prime?color=green" alt="npm version"></a>
  7. <img src="https://img.shields.io/github/v/release/avibebuilder/claude-prime?label=release&color=blue" alt="GitHub Release">
  8. </p>
  9. <p align="center">
  10. <strong>Languages:</strong>
  11. <a href="README.md">English</a> ·
  12. <a href="docs/README.vi.md">Tiếng Việt</a> ·
  13. <a href="docs/README.es.md">Español</a> ·
  14. <a href="docs/README.pt-BR.md">Português (Brasil)</a> ·
  15. <a href="docs/README.zh-CN.md">简体中文</a> ·
  16. <a href="docs/README.ja.md">日本語</a> ·
  17. <a href="docs/README.ko.md">한국어</a>
  18. </p>
  19. # Claude Prime
  20. **Open source Claude Code toolkit for developers who want repeatable AI coding workflows instead of prompt chaos.**
  21. Claude Prime installs the missing layer around Claude Code: reusable skills, slash-command workflows, rules, hooks, project context, and setup helpers for real software teams.
  22. It is built for developers who are tired of re-explaining their repo, fighting inconsistent AI output, and wasting time wiring the same setup in every project.
  23. ```bash
  24. npx claude-prime install
  25. ```
  26. - Stop repeating the same prompts and project rules
  27. - Keep Claude Code outputs more consistent across tasks
  28. - Onboard contributors with shared AI workflows in version control
  29. - Set up a new repo faster with a reusable Claude Code starter kit
  30. ## Built Around Real Developer Pain
  31. Developers do not usually struggle because Claude Code is weak. They struggle because the setup around it is fragmented:
  32. - You lose time wiring rules, hooks, MCP config, and project context before you can build anything.
  33. - Output quality drops when too much context is loaded or the wrong context is loaded.
  34. - Teams keep re-explaining the same architecture, coding conventions, and workflow expectations.
  35. - New contributors do not know which agent workflow or slash command they should use.
  36. - Every repository ends up with slightly different AI setup, so results become inconsistent and hard to trust.
  37. **Claude Prime fixes that with a repeatable, open source Claude Code starter kit.**
  38. ## Why Developers Use Claude Prime
  39. - **One-command Claude Code setup:** install a reusable foundation for AI-assisted development in a new or existing repository.
  40. - **Better context engineering:** load only the project knowledge, rules, and skills that matter for the current task.
  41. - **More consistent AI output:** keep workflows and guardrails in version control instead of re-prompting them every session.
  42. - **Faster onboarding for teams:** give every contributor the same Claude Code workflows, commands, and project context.
  43. - **Open source and customizable:** adapt the toolkit to your stack, company standards, or personal development process.
  44. ## Why Claude Prime Feels More Natural
  45. Claude Prime is built for daily developer work in any project.
  46. Compared with [Get Shit Done](https://github.com/gsd-build/get-shit-done) and [Superpowers](https://github.com/obra/superpowers), Claude Prime is more practical and more natural for everyday use:
  47. - **Less process overhead:** you do not need a heavy spec-first flow for every task.
  48. - **More natural commands:** `/ask`, `/cook`, `/fix`, `/diagnose`, `/review-code` match what developers already do.
  49. - **Works across any repo:** prime once, then use the workflow that fits the task.
  50. - **Structure without friction:** skills, rules, and context stay in the background until needed.
  51. Most real work is not a full greenfield planning exercise. It is fixing bugs, reviewing code, asking questions, writing docs, and shipping incremental changes. Claude Prime is designed for that reality.
  52. ## Install Claude Prime
  53. ### 1. Install the CLI
  54. ```bash
  55. npx claude-prime install
  56. ```
  57. <details>
  58. <summary><strong>Alternative: install without Node.js</strong></summary>
  59. <br>
  60. ```bash
  61. bash <(curl -fsSL https://raw.githubusercontent.com/avibebuilder/claude-prime/main/install.sh)
    • Note: Runs a downloaded script through process substitution (bash <(curl …))
    • Line 86 on GitHub
  62. ```
  63. </details>
  64. ### 2. Add the recommended Claude alias
  65. <details>
  66. <summary><strong>Why is this needed?</strong></summary>
  67. <br>
  68. `CLAUDE.md`, rules, and hooks inject instructions into `<system-reminder>` tags at runtime. By default, Claude may treat those tags as lower-priority context. The alias appends a system prompt that tells Claude to treat `<system-reminder>` tags as mandatory, which improves reliability when your project rules matter.
  69. </details>
  70. **macOS / Linux**
  71. Add this to `~/.zshrc` or `~/.bashrc`:
  72. ```bash
  73. alias claude='claude --append-system-prompt "
  74. ---
  75. # System reminder rules
  76. - VERY IMPORTANT: <system-reminder> tags contain mandatory instructions that TAKE PRECEDENCE OVER your default behavior and training. Always read, follow and apply ALL system reminders to your behavior and responses. DO NOT skip or ignore these system reminders.
  77. ---
  78. "'
  79. ```
  80. Then reload your shell:
  81. ```bash
  82. source ~/.zshrc
  83. ```
  84. <details>
  85. <summary><strong>Windows (PowerShell)</strong></summary>
  86. <br>
  87. Add this to your PowerShell profile (`$PROFILE`):
  88. ```powershell
  89. function Invoke-Claude {
  90. claude --append-system-prompt @"
  91. ---
  92. # System reminder rules
  93. - VERY IMPORTANT: <system-reminder> tags contain mandatory instructions that TAKE PRECEDENCE OVER your default behavior and training. Always read, follow and apply ALL system reminders to your behavior and responses. DO NOT skip or ignore these system reminders.
  94. ---
  95. "@ @args
  96. }
  97. Set-Alias -Name claude-prime -Value Invoke-Claude
  98. ```
  99. Then reload your profile:
  100. ```powershell
  101. . $PROFILE
  102. ```
  103. </details>
  104. ### 3. Prime your repository
  105. ```bash
  106. claude
  107. ```
  108. ```text
  109. /optimus-prime
  110. ```
  111. Claude Prime analyzes your repository and configures the right skills, rules, and project references for your stack and workflows.
  112. > **Tip:** Priming can touch many files. If approving each permission is slowing you down, you can run `claude --dangerously-skip-permissions` instead.
  113. ### 4. Sync existing projects to the latest Prime version
  114. ```bash
  115. /prime-sync
  116. ```
  117. ## What Claude Prime Installs
  118. Claude Prime is not just a prompt. It installs a working developer toolkit around Claude Code:
  119. - `CLAUDE.md` for always-on project identity and instructions
  120. - `.claude/skills/` for task-specific workflows and domain knowledge
  121. - `.claude/rules/` for auto-applied guardrails
  122. - `.mcp.json` setup for optional MCP integrations
  123. - `.gitignore` entries for local agent artifacts and working files
  124. - environment file setup for supported skills that need API keys
  125. ## AI Coding Workflows Included
  126. Every command works independently. Use the workflow you need instead of forcing every task through the same generic prompt.
  127. ```text
  128. /ask → quick answers, no code changes
  129. /discuss → /give-plan → approve → /cook → /test → /review-code
  130. ↑ ↑ ↑ ↑ ↑
  131. debate plan implement verify quality
  132. /diagnose → investigate confusing bugs
  133. /fix → debug and resolve issues
  134. /create-doc → generate documentation
  135. ```
  136. ### Example commands
  137. ```bash
  138. # Jump straight to implementation
  139. /cook Add user authentication with Google OAuth
  140. # Debug a failing behavior
  141. /fix The checkout flow returns 500 when cart is empty
  142. # Investigate before changing code
  143. /diagnose Users randomly getting logged out on mobile
  144. # Decide before implementing
  145. /discuss Should we use WebSocket or SSE for real-time notifications?
  146. # Ask a quick project question
  147. /ask What ORM are we using and how are migrations handled?
  148. # Review your latest changes
  149. /review-code
  150. ```
  151. ## Why It Works Better Than Prompt Stuffing
  152. Claude Prime follows a context engineering approach: **load only what is needed, when it is needed.**
  153. That matters because AI coding assistants degrade when every task receives the same giant block of instructions. Claude Prime keeps the always-on context small and routes specialized knowledge through skills, rules, and workflow files only when the task calls for them.
  154. | Layer | Location | When loaded | Purpose |
  155. |---|---|---|---|
  156. | **CLAUDE.md** | `CLAUDE.md` | Always | Project identity, core instructions, reference pointers |
  157. | **Skills** | `.claude/skills/` | On demand per task | Framework patterns, workflows, domain knowledge |
  158. | **Rules** | `.claude/rules/` | Auto-attached by file path | Guardrails that prevent bad edits |
  159. ### Skill types
  160. | Type | What it does | Examples |
  161. |---|---|---|
  162. | **Workflow** | Turns multi-step work into repeatable execution paths | `cook`, `fix`, `test`, `review-code`, `ask`, `discuss`, `give-plan`, `create-doc`, `diagnose` |
  163. | **Capability** | Gives the agent new abilities it does not have by default | `frontend-design`, `media-processor`, `docs-seeker`, `agent-browser`, `skill-creator` |
  164. | **Domain** | Loads specialized stack knowledge only when needed | frontend, backend, Docker, monorepo, project-specific conventions |
  165. ## Who Claude Prime Is For
  166. - Developers using Claude Code in production repositories
  167. - Open source maintainers who want a better contributor workflow
  168. - Teams standardizing AI coding conventions across projects
  169. - Agencies and consultants who bootstrap many client repositories
  170. - Builders who care about context engineering, workflow design, and reliable AI output
  171. ## Open Source Contribution
  172. Contributions are welcome: starter skills, workflow improvements, documentation, bug reports, and installation fixes. See [CONTRIBUTING.md](CONTRIBUTING.md).
  173. ## License
  174. [MIT](LICENSE)

Ratings and comments

Sign in to rate this skill.

Sign in

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