Clawdstrike

ClawdStrike Claude Code Plugin

Runtime security enforcement for AI coding agents. This plugin integrates ClawdStrike's policy engine, threat hunting, and audit system directly into Claude Code.

What It Does

• Pre-tool policy checks -- every tool call is evaluated against your security policy before execution
• Post-tool audit receipts -- signed attestations of every action for compliance and forensics
• 10 MCP tools -- security scanning, threat hunting, event correlation, and policy management
• 3 auto-triggering skills -- contextual security guidance that activates when relevant
• 5 slash commands -- quick access to scanning, auditing, and posture assessment
• 1 specialist agent -- deep security review with OWASP and policy compliance checks

Installation

From the marketplace (recommended)

# Add the ClawdStrike marketplace
/plugin marketplace add backbay-labs/clawdstrike

# Install the plugin
/plugin install clawdstrike@clawdstrike

From a local clone

claude --plugin-dir ./clawdstrike-plugin

Quick Reference

Slash Commands

Command	Description
/clawdstrike:scan	Scan MCP server configurations for security issues
/clawdstrike:audit	Show security audit trail for the current session
/clawdstrike:posture	Assess overall security posture (A-F grade)
/clawdstrike:policy	Show active security policy and guard details
/clawdstrike:tui	Launch the ClawdStrike TUI dashboard

MCP Tools

Tool	Description
clawdstrike_check	Evaluate an action against the active policy
clawdstrike_scan	Scan MCP server configurations
clawdstrike_query	Query security events with filters
clawdstrike_timeline	Get chronological event timeline
clawdstrike_correlate	Correlate events to detect attack patterns
clawdstrike_ioc	Check indicators of compromise
clawdstrike_policy_show	Show policy and guard configuration
clawdstrike_policy_eval	Evaluate a hypothetical action against policy
clawdstrike_hunt_diff	Diff security state between two points in time
clawdstrike_report	Generate a structured security report

Skills (Auto-Triggering)

Skill	Triggers On
Security Review	Edits to sensitive paths, shell commands, dependency changes
Threat Hunt	Security events, suspicious activity, IOC investigation
Policy Guide	Questions about what's allowed, guard behavior, rulesets

Agent

Agent	Purpose
security-reviewer	Deep code security review with OWASP checks and policy verification

Architecture

claude code
    |
    v
+-------------------+
| clawdstrike-plugin|
+-------------------+
|                   |
| hooks/            |--- pre-tool-check.sh ----> ClawdStrike CLI
|   hooks.json      |<-- allow/deny verdict ----/
|                   |
|                   |--- post-tool-receipt.sh -> ClawdStrike CLI
|                   |<-- signed receipt --------/
|                   |
| scripts/          |--- session-start.sh ----> Initialize session
|   mcp-server.ts   |--- session-end.sh ------> Finalize + report
|                   |
| skills/           |--- security-review/  (auto-trigger on risky edits)
|                   |--- threat-hunt/      (auto-trigger on investigations)
|                   |--- policy-guide/     (auto-trigger on policy questions)
|                   |
| commands/         |--- /scan     (MCP config scanning)
|                   |--- /audit    (session audit trail)
|                   |--- /posture  (security posture grade)
|                   |--- /policy   (active policy details)
|                   |--- /tui      (interactive dashboard)
|                   |
| agents/           |--- security-reviewer (deep code review agent)
+-------------------+
         |
         v
  MCP Server (stdio)
  10 tools via @modelcontextprotocol/sdk
         |
         v
  ClawdStrike CLI / API

Configuration

Environment Variables

Variable	Description	Default
CLAWDSTRIKE_ENDPOINT	ClawdStrike API endpoint (CLI path or HTTP URL)	clawdstrike (PATH lookup)
CLAWDSTRIKE_SESSION_ID	Session identifier for audit trail continuity	Auto-generated UUID
CLAWDSTRIKE_HOOK_FAIL_OPEN	Allow actions when ClawdStrike is unavailable. Accepts true, 1, or yes	false (fail-closed)

Fail-Closed by Default

The plugin follows ClawdStrike's fail-closed design philosophy:

• If the ClawdStrike CLI is unavailable, tool calls are blocked (unless CLAWDSTRIKE_HOOK_FAIL_OPEN=true)
• If a policy fails to parse, the session refuses to start
• If a guard errors during evaluation, the verdict is deny

Set CLAWDSTRIKE_HOOK_FAIL_OPEN=true only in development environments where security enforcement is not required.

Development

# Install dependencies
bun install

# Run MCP server standalone
bun run mcp

# Type check
bun run typecheck

# Run tests
bun test

License

MIT