Best practices for applying Amazon Bedrock Guardrails to code generation workflows

<p><em>This post continues our series on best practices with Amazon Bedrock Guardrails. For the previous post, see <a href="https://aws.amazon.com/blogs/machine-learning/build-safe-generative-ai-applications-like-a-pro-best-practices-with-amazon-bedrock-guardrails/" target="_blank" rel="noopener">Build safe generative AI applications like a pro: best practices with Amazon Bedrock Guardrails</a>.</em></p> <p>AI-powered coding assistants and code generation workflows, such as Claude Code, <a href="https://aws.amazon.com/documentation-overview/kiro/" target="_blank" rel="noopener">Kiro</a>, and <a href="https://chatgpt.com/codex" target="_blank" rel="noopener">OpenAI Codex</a>, are transforming how developers write software. These tools generate code in real time through streaming responses, often producing thousands of characters across extended sessions. As organizations adopt generative AI workflows with code at scale using these assistants, it is important that unsafe code patterns are detected and blocked whenever required. <a href="https://aws.amazon.com/bedrock/guardrails/" target="_blank" rel="noopener">Amazon Bedrock Guardrails</a> helps detect and filter unsafe and undesired code content with safeguards such as content filters for content moderation, prompt attack prevention with jailbreaks, prompt injection, and prompt leakage, sensitive information filters to redact and block personally identifiable information (PII), and more.</p> <p>However, coding workflows along with agentic loops have unique throughput characteristics including long streaming outputs, concurrent developer sessions, and repetitive context evaluation. Applying Amazon Bedrock Guardrails to workflows with these characteristics without proper configuration could potentially lead to constraints such as throttling errors, increased costs, and less than optimal latency. In this post, we explain how Amazon Bedrock Guardrails can be configured for code generation workflows with coding assistants to overcome these constraints. With these best practices, you can build an efficient blueprint helping you with effective capacity planning with robust safety coverage.</p> <h2 id="why-guardrails-are-important-for-code-generation-workflows">Why guardrails are important for code generation workflows</h2> <p>Amazon Bedrock Guardrails offers comprehensive safeguards that detect and filter harmful and undesirable content from both user inputs and model responses, to help build safe generative AI applications. While these safeguards can be configured and implemented across a variety of applications, there are specific safeguards that can be used to protect harmful code patterns such as:</p> <ul> <li><strong>Detect prompt attacks</strong> – block attempts to manipulate the model into generating malicious code or bypassing instructions.</li> <li><strong>Filter sensitive information</strong> – can help block or mask sensitive information such as personally identifiable information (PII), custom regex in user inputs and model responses, catch hardcoded AWS access keys, database connection strings, or private keys before they appear in generated code.</li> <li><strong>Content moderation</strong> – detect and filter harmful content in user prompts and model responses, including content that violates your organization’s acceptable use policies.</li> <li><strong>Block denied topics</strong> – define custom topics that are used as a basis for blocking conversations related to prohibited activities, such as generating code to bypass authentication mechanisms or assisting with unauthorized access patterns.</li> </ul> <p>These safeguards are essential when AI-generated code is used in production systems.</p> <h3 id="a-scenario-when-good-guardrails-go-sideways">A scenario: When good guardrails go sideways</h3> <p>A customer system team has just rolled out Claude Code on Amazon Bedrock to 15 developers. They’ve configured a guardrail with three safeguards: prompt att