skills.rest
arcjet
arcjetOFFICIAL
Explore all skills by this author →
Website𝕏@arcjetGitHub

integrate-arcjet-guard-claude-agent-sdk-py

Integrates Arcjet Guard security controls into Python Claude Agent SDK tools and hooks.

1|Updated Apr 8, 2026
Search Tags:#rate limiting#python#ai security#prompt injection#claude-agent-sdk#arcjet#tool guard
One-click install
npx skills add https://github.com/arcjet/arcjet-plugin --skill integrate-arcjet-guard-claude-agent-sdk-py-arcjet
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: integrate-arcjet-guard-claude-agent-sdk-py
Source: https://github.com/arcjet/arcjet-plugin/tree/main/plugins/arcjet/skills/integrate-arcjet-guard-claude-agent-sdk-py
Command: npx skills add https://github.com/arcjet/arcjet-plugin --skill integrate-arcjet-guard-claude-agent-sdk-py-arcjet

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires arcjet, claude-agent-sdk.

What problem does it solve? AI agents built with the Python Claude Agent SDK can execute risky tool calls, process prompt-injected input, and exceed rate limits without any policy enforcement. This Skill adds Arcjet Guard protection to those agents so tool invocations, inbound prompts, and built-in or MCP tool calls are screened and rate-limited before damage occurs. ## Core Features & Use Cases - Authored tool gating: Wraps @tool definitions with guard_tool so denied calls return JSON-in-content with is_error: True and the handler never runs. - Inbound prompt screening: Uses guard_hooks on UserPromptSubmit to block prompt injection before the model reads the prompt. - Built-in and MCP tool control: Applies PreToolUse hooks with permissionDecision: deny for unwrapped built-ins and MCP tools, with an exclude list to avoid double-guarding wrapped tools. - Use Case: A support agent with a lookup_order tool gets a per-user token bucket rate limit, inbound prompt injection detection, and fail-closed behavior when the Arcjet guard is unreachable. ## Quick Start Ask your agent to add Arcjet Guard to the Python claude-agent-sdk project by wrapping the authored tools with guard_tool and adding guard_hooks for inbound prompts and unwrapped tools.

Frequently Asked Questions about integrate-arcjet-guard-claude-agent-sdk-py

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I add Arcjet Guard to the Python Claude Agent SDK?▼

Install arcjet with the claude-agent-sdk extra, launch a guard client once at module scope with launch_arcjet, then wrap authored @tool functions with guard_tool and pass guard_hooks into ClaudeAgentOptions for inbound and PreToolUse enforcement.

How do I rate limit tools in claude-agent-sdk?▼

Create a TokenBucket rule with a label, bucket, refill rate, and max tokens, then pass it to guard_tool for authored tools or to guard_hooks for built-ins and MCP tools. Key the bucket on the authenticated user ID, not a model-supplied value.

Does this work with the JavaScript claude-agent-sdk or Claude Managed Agents?▼

No. This module targets only the official Python claude-agent-sdk version 0.2.127 or later. The JS adapter @arcjet/guard/claude-agent-sdk/v0 and arcjet.guard.claude_managed_agents are separate integrations with different APIs.

Why is my guarded tool being called twice by Arcjet?▼

Double guard decisions happen when a guard_tool-wrapped tool is missing from the exclude list in guard_hooks. Add each wrapped tool as a server/name entry in exclude so PreToolUse does not evaluate it a second time.

What happens when the Arcjet guard is unreachable?▼

Every helper defaults to on_guard_error="deny", so tool calls and inbound prompts fail closed. For UserPromptSubmit specifically, choosing "allow" is a legitimate option if blocking all agent responses during an outage is unacceptable.

Related Skills

Explore More Software Engineering Skills→
faw01faw01
18

arcjet

Implement rate limiting, bot detection, and prompt injection scanning via the Arcjet SDK.

Community
Advanced
davila7davila7
30.5k

Wireshark Network Traffic Analysis

Capture, filter, and analyze network packets with Wireshark for security investigations and troubleshooting.

Community
Intermediate
olostepolostep

integrate

Automate Olostep SDK integration with idiomatic code generation for supported frameworks.

Community
Advanced
MapleTechLabsMapleTechLabs
1.6k

maple-java-style

Configure Java applications to export OpenTelemetry traces, logs, and metrics to Maple.

Official
Intermediate