TeamAI — Claude Code Plugin for Dev Teams

A comprehensive Claude Code plugin that gives your team standardized code review, security scanning, PR workflows, and agent orchestration — out of the box.

What's Included

Slash Commands

| Command | Description | |---------|-------------| | /project:TeamAI:CodeReview | Multi-agent code review (architecture, DRY, security, performance) | | /project:TeamAI:PRReady | Pre-PR audit: lint, build, tests, secrets scan, PR summary generation | | /project:keys-scan | Scan codebase for leaked secrets, API keys, tokens, credentials | | /project:readme-update | Analyze recent changes and suggest documentation updates | | /project:check-standards | Audit files against the team style guide | | /project:openspec-setup | Setup guide and workflow for OpenSpec spec-first development |

Agents

Specialized agents spawned by commands or manually via agent teams:

• Backend — APIs, databases, server-side architecture
• Frontend — UI components, styling, accessibility, state management
• Planning — Task decomposition, architecture decisions, risk assessment
• Security — Vulnerability scanning, OWASP, secrets detection
• PR Review — Pull request review with actionable feedback
• Standards Enforcer — Style guide compliance, naming, structure

Rules (auto-loaded)

• code-quality.md — DRY, no hardcoded values, guard at source, single responsibility
• security.md — Secrets, injection prevention, auth, data protection
• agent-orchestration.md — How to use agent teams effectively

Integrations & MCPs

| Integration | Type | Config needed? | What it does | |-------------|------|---------------|-------------| | Context7 | MCP server (npm) | In settings.json ✅ | Up-to-date docs for any library — Claude fetches live documentation instead of relying on training data | | Playwright | MCP server (npm) | In settings.json ✅ | Browser automation, E2E testing, screenshot capture directly from Claude | | Figma | Built-in cloud MCP | None — already in Claude Code | Design-to-code, code-to-design, component inspection, asset export | | Atlassian | Built-in cloud MCP | None — already in Claude Code | Read/create Jira issues, read/edit Confluence pages, search across projects | | OpenSpec | CLI tool (not an MCP) | npm install -g + openspec init | Spec-first workflow: proposals, specs, design docs, task checklists before code | | CodeRabbit | GitHub/GitLab bot | Install app + .coderabbit.yaml | AI-powered PR reviews on every push (complements local Claude review) |

Why some are MCPs and some aren't:

• MCPs (Context7, Playwright) run as local servers that Claude Code talks to — configured in settings.json
• Built-in cloud MCPs (Figma, Atlassian) are provided by Claude Code itself — no config, just authenticate when prompted
• OpenSpec is a CLI that generates files and registers its own /opsx:* commands into .claude/commands/
• CodeRabbit is a separate GitHub bot — it reviews PRs independently, not through Claude Code

Git Hooks

• pre-commit — secrets scan, lint, format check, build verification (auto-detects your stack)

Style Guide

• standards/style-guide.md — naming, structure, error handling, testing conventions

Installation

teamai init (recommended)

npx teamai-plugin init

That's it. Run it inside your project directory and it:

1. Scans your project — detects language, framework, linter, formatter, CI/CD, package manager
2. Generates a CLAUDE.md pre-filled with your detected tech stack and common commands
3. Customizes .claude/settings.json with permissions tailored to your tools
4. Installs commands, agents, rules, style guide, git hooks, OpenSpec, CodeRabbit config
5. Preserves your existing files — never overwrites what you already have

Other install methods

# Install globally for repeated use across projects
npm install -g teamai-plugin
teamai init                          # smart init in current directory
teamai init ./my-app                 # smart init in specific directory
teamai install                       # full install (no scan, raw template copy)

# One-liner from GitHub (no npm)
curl -fsSL https://raw.githubusercontent.com/roytol-abra/team-ai-plugin/main/install.sh | bash

# Private repo
gh repo clone roytol-abra/team-ai-plugin /tmp/team-ai-plugin && /tmp/team-ai-plugin/setup.sh .

CLI Commands

teamai init [path]       # Scan project + smart install (recommended)
teamai install [path]    # Full install without scanning (raw template)
teamai update [path]     # Update existing installation (preserves customizations)
teamai doctor [path]     # Check installation health and integration status
teamai --help            # Show help

init vs install

| | teamai init | teamai install | |---|---|---| | Scans project | ✅ Detects stack, tools, CI/CD | ❌ No scanning | | CLAUDE.md | Auto-filled with detected stack | Generic template | | settings.json | Permissions customized for your tools | Generic permissions | | Existing files | Merges (adds missing MCPs, preserves yours) | Skips existing | | Best for | New project setup | Quick copy of raw templates |

Manual Install

Copy these into your project root:

1. .claude/ directory (commands, agents, rules, settings)
2. CLAUDE.md
3. standards/style-guide.md
4. .coderabbit.yaml
5. hooks/pre-commit → .git/hooks/pre-commit (and chmod +x)

Then install OpenSpec: npm install -g @fission-ai/openspec@latest && openspec init

What the installer does automatically

| Step | What | Auto? | |------|------|-------| | 1 | Copies .claude/ config (commands, agents, rules, settings) | ✅ | | 2 | Copies CLAUDE.md template | ✅ | | 3 | Copies standards/style-guide.md | ✅ | | 4 | Installs git pre-commit hook | ✅ | | 5 | Installs OpenSpec CLI globally + initializes in project | ✅ | | 6 | Copies .coderabbit.yaml config | ✅ | | 7 | Context7 + Playwright MCPs (download on first Claude Code launch) | ✅ | | 8 | Figma + Atlassian (built-in, authenticate on first use) | ✅ | | 9 | CodeRabbit GitHub app (requires org admin) | ❌ Manual |

After Installation

Required (2 minutes):

1. Edit CLAUDE.md — fill in your project overview, tech stack, and architecture
2. Open Claude Code and run /project:TeamAI:CodeReview to verify everything works

Optional: 3. Install CodeRabbit app at https://coderabbit.ai (needs org admin) 4. Authenticate Figma and Atlassian when prompted on first use in Claude Code 5. Configure OpenSpec profile: openspec config profile

Customization

Adding Project-Specific Rules

Add .md files to .claude/rules/ — they're automatically loaded by Claude Code.

Adding Custom Commands

Add .md files to .claude/commands/ — they become /project:command-name slash commands.

Modifying Agents

Edit files in .claude/agents/ to adjust agent behavior, expertise, or output format.

Overriding Standards

Project-specific rules in CLAUDE.md always override the style guide defaults.

File Structure

.claude/
  commands/
    TeamAI:CodeReview.md    — multi-agent code review
    TeamAI:PRReady.md       — PR readiness check
    keys-scan.md            — secrets scanner
    readme-update.md        — README update suggestions
    check-standards.md      — style guide audit
    openspec-setup.md       — OpenSpec workflow guide
  agents/
    backend.md              — backend specialist agent
    frontend.md             — frontend specialist agent
    planning.md             — planning/architecture agent
    security.md             — security specialist agent
    pr-review.md            — PR review agent
    standards-enforcer.md   — standards enforcement agent
  rules/
    code-quality.md         — code quality rules
    security.md             — security rules
    agent-orchestration.md  — agent team coordination rules
  settings.json             — MCP servers, permissions, hooks
CLAUDE.md                   — project config template
standards/
  style-guide.md            — team style guide
hooks/
  pre-commit                — git pre-commit hook
setup.sh                    — installer script