Claude Code in CI/CD without handing it the keys
Headless Claude Code reviewing pull requests and fixing failing builds in GitHub Actions against Amazon Bedrock, with a trust boundary untrusted PR content can't cross and merging left to people.
architectures
Reference designs and decision records for Claude systems in production.
4 architectures published
All architectures
Document extraction that knows when to ask a person
An event-driven pipeline that extracts structured data from documents with Claude on Amazon Bedrock, checks it against evidence, and sends only the doubtful cases to human review.
A multi-agent research pipeline, and when to use one agent instead
An orchestrator that splits a research question across parallel workers with narrow contexts, then synthesises cited findings and names what it couldn't cover.
A support agent on AgentCore that can't give away the store
A customer support agent with account-changing tools, where authorisation lives in code and policy rather than the prompt, and hard cases reach a person with context attached.
Nothing in this pillar yet.