Documentation

Everything you need to secure your AI agents with SpiderShield.

Quickstart

Python SDK

python

pip install spidershield

from spidershield import SpiderGuard, Decision

guard = SpiderGuard(policy="balanced", dlp="redact")

# Check before tool execution
result = guard.check("execute_sql", {"query": "SELECT * FROM users"})
if result.decision == Decision.ALLOW:
    output = run_tool("execute_sql", {"query": "SELECT * FROM users"})
    # Scan output for secrets / PII
    clean = guard.after_check("execute_sql", output)

CLI

bash

# Guard mode — wrap any MCP server
"code-keyword">$ "code-function">spidershield guard "code-keyword">--preset balanced -- "code-function">npx @modelcontextprotocol/server-filesystem /tmp

# Scan mode — static security analysis
"code-keyword">$ "code-function">spidershield scan ./my-mcp-server

# Proxy mode — transparent interception
"code-keyword">$ "code-function">spidershield proxy "code-keyword">--policy strict -- python my_mcp_server.py

Guides

Runtime Guard

Policy enforcement before every tool call. ALLOW, DENY, or ESCALATE decisions in real time.

DLP Scanner

Detect and redact PII, API keys, secrets, and prompt injection in tool outputs.

CLI Reference

Full command reference for scan, guard, proxy, audit, policy, and dataset commands.

Policy Engine

YAML-based policy presets (permissive, balanced, strict) and custom rule authoring.

Audit Logging

JSONL audit trail for every tool call, decision, and DLP event. Queryable via CLI.

Agent Security

Config audit, skill scanning, toxic flow detection, and content pinning for agent frameworks.

Installation

PyPI

pip install spidershield

pipx (CLI only)

pipx install spidershield

From source

git clone && pip install -e .

View on GitHub