eneoplugin

A Claude Code plugin marketplace with AI team behavioral guidelines for Eneo developers.

Installation

1. Add the marketplace

/plugin marketplace add CCimen/eneoplugin

2. Install plugins

/plugin install karpathy-guidelines@eneoplugin
/plugin install frontend-design@eneoplugin
/plugin install checker@eneoplugin
/plugin install vikunja-kanban@eneoplugin
/plugin install github-findings@eneoplugin
/plugin install codex-review@eneoplugin
/plugin install gemini-review@eneoplugin

Available Plugins

karpathy-guidelines

Behavioral guidelines to reduce common LLM coding mistakes. Includes:

• CLAUDE.md - Automatically loaded guidelines emphasizing:

  1. Think Before Coding
  2. Simplicity First
  3. Surgical Changes
  4. Goal-Driven Execution

• Skill - Invoke with /karpathy-guidelines for detailed guidance when writing, reviewing, or refactoring code.

frontend-design

Create distinctive, production-grade frontend interfaces with high design quality. Includes:

• Skill - Invoke with /frontend-design when building web components, pages, dashboards, or any web UI.

Focuses on bold aesthetic choices, avoiding generic "AI slop" aesthetics. Emphasizes typography, color themes, motion, spatial composition, and visual details.

checker

Automated Pyright type checking for the eneo Python backend. Includes:

• Stop Hook - Automatically runs type checking when Claude finishes editing Python files in backend/src/intric/. Blocks and provides feedback if errors are found.

• CLAUDE.md - Guidelines for the ratcheting type checking strategy.

• Skill - Invoke with /checker to manually run type checks.

Configuration:

Environment Variable	Effect
TYPECHECK_DISABLE=1	Completely disable type checking
TYPECHECK_WARN_ONLY=1	Show errors but don't block

github-findings

Track findings, bugs, and ideas as GitHub issues on the eneo project board. Designed for discovering things mid-conversation that should be tracked but not fixed right now.

• Skill - Invoke with /finding to create, list, or grab issues.

Command	Description
/finding	Create a new issue from a finding in the current conversation
/finding list	List all open items (Todo + In Progress) on the board
/finding grab <number>	Pick up an issue — assigns you, sets "In Progress", creates a branch

Issues are written with structured sections (Problem, Context, Suggested Solution, Acceptance Criteria) so that Claude or another AI can understand and work on them later without extra context.

Setup (required once per developer):

1. Install the plugin: /plugin install github-findings@eneoplugin
2. Add GitHub project scope to your token:

   gh auth refresh -s project -h github.com

   This opens a browser where you authorize the extra scope. Only needed once.
3. Verify it works:

   gh project view 1 --owner eneo-ai

Note: The skill targets the eneo-ai/eneo repository and project board #1. All team members with repo access and the project token scope can use it.

vikunja-kanban

Create and update Vikunja Kanban cards with safe, high-level Swedish progress updates. Includes:

• Skill - Invoke with /vikunja-kanban to create cards, link PRs, post progress updates, move tasks, and manage labels.

Configuration:

Environment Variable	Effect
VIKUNJA_BASE_URL	Vikunja server root (no /api/v1)
VIKUNJA_API_TOKEN	API token (Bearer token)
VIKUNJA_PROJECT_NAME	Default project name (default: Internal TODO)
VIKUNJA_VIEW_NAME	Default view name (default: Kanban)

codex-review

Always-on peer review via OpenAI Codex CLI. Claude automatically invokes this for plan reviews, mid-implementation code reviews, and consultations. Includes:

• Skill - Claude auto-invokes or use /codex-review manually. Three operating modes:

  • Plan Review — Verifies architecture before implementation (~3 iterations, gpt-5.3-codex with xhigh reasoning)
  • Code Review — Reviews git diff mid-implementation for quality, maintainability, testability (gpt-5.3-codex-spark with xhigh)
  • Consultation — Answers questions when Claude is unsure (gpt-5.3-codex-spark with xhigh)

• Four Personas — Solution Architect (always), API Consumer (endpoints), Security Reviewer (auth/permissions), Performance Analyst (queries/hot paths)

• Eneo Context — Includes architecture reference so Codex understands Eneo patterns

Requires: Codex CLI installed (codex --version)

gemini-review

Architectural cross-verification via Google Gemini CLI. Claude invokes this AFTER codex-review to get an independent third perspective. Includes:

• Skill - Claude auto-invokes after codex-review or use /gemini-review manually. Operates as the third step in the review pipeline:

  Claude → Codex (primary review) → Gemini (cross-check) → Claude (synthesis)
  

• Same Four Personas — Solution Architect, API Consumer, Security Reviewer, Performance Analyst

• Focus — Catches blind spots that Claude and Codex may share. Provides independent architectural verification using gemini-3.1-pro-preview

Requires: Gemini CLI installed (gemini --version)

Usage

Once installed, CLAUDE.md guidelines are automatically applied. Invoke skills manually:

/karpathy-guidelines    # Coding guidelines
/frontend-design        # UI design guidance
/checker                # Run type checks
/finding                # Create/list/grab GitHub project issues
/vikunja-kanban         # Vikunja Kanban card management
/codex-review           # Peer review via Codex CLI
/gemini-review          # Cross-check via Gemini CLI

License

MIT