Cyera Inline Protection Plugin for Claude Code

Intercepts Claude Code tool calls and user prompts for policy enforcement.

Installation

Option A: User self-install

# Add the marketplace
/plugin marketplace add cyera-ai/inline-protection-plugin

# Install the plugin
/plugin install cyera-inline-protection@cyera-security

Option B: Admin force-deploy (managed settings)

Add to managed settings (via admin console, MDM, or managed-settings.json):

{
  "extraKnownMarketplaces": {
    "cyera-security": {
      "source": { "source": "github", "repo": "cyera-ai/inline-protection-plugin" }
    }
  },
  "enabledPlugins": { "cyera-inline-protection@cyera-security": true },
  "allowManagedHooksOnly": true
}

Option C: Local testing

claude --plugin-dir ./plugins/cyera-inline-protection

What it does

Registers hooks on:

• PreToolUse (Write, Edit, Bash) — before tool execution
• PostToolUse (Write, Edit, Bash) — after tool execution
• UserPromptSubmit — before Claude processes user input

Currently logs all events to ~/.claude/cyera-events.log for observation.

Log format

---
timestamp: 2026-06-21T10:30:00Z
event: PreToolUse
tool: Bash
session: abc123
cwd: /Users/dev/project
payload:
  { full JSON payload }

Roadmap

• v0.1.0: Logging only (current)
• v0.2.0: HTTP callout to Cyera policy engine
• v0.3.0: Blocking based on policy response
• v1.0.0: Production-ready with auth, config, and error handling