zeude — Enterprise AI Adoption Platform for Claude Code

Measure usage. Sync config. Guide teams. Turn AI investment into AI fluency.

zeude

Installation

Plugin (Claude Code)

claude plugin marketplace add zep-us/zeude

Self-hosted (recommended for enterprise)

git clone https://github.com/zep-us/zeude.git
cd zeude
cp .env.example .env
# Configure Supabase and ClickHouse credentials in .env

# Start dashboard
cd zeude/dashboard && pnpm install && pnpm dev

Skill (any platform)

npx skills add https://github.com/akillness/oh-my-skills --skill zeude

Requirements

• Claude Code installed and authenticated
• Supabase account — configuration storage and user management
• ClickHouse instance — analytics backend
• macOS or Linux (x86_64 / arm64)
• Go (for building the CLI shim)
• Node.js + pnpm (for the dashboard)

When to use

• Measure Claude Code adoption rates across an engineering org without mandating behavior
• Centrally deploy skills, MCP servers, and hooks to all developers in one push
• Surface context-aware skill suggestions at the moment a developer types a prompt
• Identify power users and surface their best practices to the rest of the team
• Run OpenTelemetry traces on Claude Code sessions for usage analytics in ClickHouse

Do not use when

• You need a single-developer setup → use standard

  skills add

  directly
• You don't have Supabase + ClickHouse infrastructure → prerequisites block self-hosting
• You need a fully managed SaaS → contact enterprise sales at jaegyu.lee@zep.us

Architecture: Three-Layer Ecosystem

Sensing   → OpenTelemetry traces → ClickHouse → Dashboard analytics
Delivery  → Dashboard API → Zeude Shim → Local ~/.claude/ configuration
Guidance  → User prompts → Hook analysis → Context-aware skill suggestions

1. Sensing — Measure adoption

# Configure OTEL endpoint in .env
OTEL_EXPORTER_OTLP_ENDPOINT=http://localhost:4318
CLICKHOUSE_DSN=clickhouse://...

• Adoption rates by team and individual
• Token usage patterns
• Most-used skills and prompts
• Power user identification

2. Delivery — Sync configuration

claude

# Per-developer install (using agent key from dashboard)
curl -fsSL https://raw.githubusercontent.com/zep-us/zeude/main/install.sh | bash -s -- --key <AGENT_KEY>

• ~/.claude/skills/

  — deployed skills
• MCP server configurations
• Hooks (UserPromptSubmit, etc.)

3. Guidance — Skill suggestions

UserPromptSubmit

# Hook fires on every Claude prompt
# Zeude analyzes intent → suggests relevant skills inline

Dashboard API

# Start dashboard (after .env configuration)
cd dashboard && pnpm dev

# Available endpoints (via dashboard)
GET  /api/skills          # list deployed skills
POST /api/skills          # deploy a skill to all team members
GET  /api/analytics       # adoption metrics
POST /api/teams           # create/manage teams

Operating rules

1. Set up Supabase and ClickHouse before any developer installs the Shim
2. Deploy the dashboard and verify API health before distributing agent keys
3. Use the Zeude Shim installer (not manual symlinks) so auto-update works
4. Monitor the ClickHouse dashboard weekly to identify adoption gaps
5. Push new skills centrally from the dashboard rather than asking developers to install manually
6. Keep agent keys per-team for access control granularity

References

• zep-us/zeude
• Docs
• Enterprise contact: jaegyu.lee@zep.us