Adrian CC - Claude Code Security Plugin

Runtime security plugin for Claude Code. It monitors every tool call, classifies it with the Adrian backend (LLM-based MAD code classification), and blocks malicious actions in real time.

Architecture

Claude Code ──► Plugin Hook ──► Adrian Backend (WebSocket, protobuf)
                   │                    │
                   │                    ├─ Classify (M0/M2/M3/M4)
                   │                    ├─ Store event + verdict
                   │                    └─ Return verdict
                   │
                   ├─ allow  (benign / policy off)
                   ├─ ask    (HITL: prompt you to approve)
                   └─ deny   (block before the tool runs)

Requirements

• Claude Code
• Python 3.12+ on your PATH (python3, python, or the py launcher on Windows)
• Windows only: Git for Windows (Git Bash). Claude Code runs plugin hooks under Git Bash by default; a .cmd launcher is bundled as a cmd.exe fallback.
• An Adrian backend and API key (Adrian Cloud, or a self-hosted OSS backend). /adrian-init walks you through this.

Install

From inside Claude Code:

/plugin marketplace add secureagentics/Adrian
/plugin install adrian-cc@adrian
/adrian-init

1. Add the marketplace - registers the adrian marketplace from the Adrian repo.
2. Install the plugin - adrian-cc from the adrian marketplace.
3. Configure - /adrian-init guides you through picking a backend (Adrian Cloud, self-hosted, or a custom URL), writes ~/.adrian/.env, asks you to drop your API key into that file (never pasted into the chat), and verifies the connection.

Monitoring is active for new Claude Code sessions once /adrian-init reports OK.

Updating

Bump happens through the marketplace. To pull the latest:

/plugin marketplace update adrian

Commands

Command	Description
/adrian-init	Guided setup: choose backend, write ~/.adrian/.env, verify the connection
/adrian-status	Show the configured backend URL, key presence, and connection health

Enforcement Modes

Enforcement is server-driven: the backend's policy decides the mode on login, and the plugin acts accordingly. No behavior is hard-wired in the client.

Mode	Behavior
ALERT	Log only. Never blocks.
BLOCK	Deny in-scope high-risk tool calls before they execute.
HITL	Prompt you in Claude Code to approve or deny the tool call inline.

Features

• Real-time blocking - high-risk actions are denied before execution
• Inline HITL - approve or deny flagged tool calls right in Claude Code
• Sub-agent tracking - tracks Agent tool spawns with parent/child hierarchy
• Tool output capture - PostToolUse logs complete tool responses
• Invocation tracking - groups events by user prompt for context
• Transcript parsing - extracts Claude's reasoning and user instructions
• Dashboard - view events and verdicts at your Adrian dashboard
• Fail-open - if the backend is unavailable, tools are allowed by default

Configuration

/adrian-init writes ~/.adrian/.env. Values already set in the environment win; otherwise a project-local .env (in the directory you launch Claude Code from) takes precedence over ~/.adrian/.env.

Variable	Default	Description
ADRIAN_API_KEY	(required)	API key for backend auth
ADRIAN_WS_URL	ws://localhost:8080/ws	Backend WebSocket URL (wss:// for Adrian Cloud)
ADRIAN_CC_VERDICT_TIMEOUT	15	Seconds to wait for a verdict
ADRIAN_CC_FAIL_OPEN	true	Allow (true) or block (false) on timeout/error

Local / development install

To run from a checkout without the marketplace (deps are still vendored, so no pip step):

claude --plugin-dir /path/to/Adrian/integrations/claude-code

Then run /adrian-init, or export ADRIAN_API_KEY and ADRIAN_WS_URL yourself.

License

Apache-2.0 - See Adrian