aws-agentcorelisted

# AWS AgentCore ## Purpose Build and operate Amazon Bedrock AgentCore agents without stuffing runtime, harness, Memory, Gateway, Identity, Observability, Browser, Code Interpreter, policy, and environment/skills details into every prompt. ## When to use Use this skill when the user asks to: - create or adapt an AgentCore project, - configure local development, invocation, packaging, deployment, runtime settings, harness settings, or environment/skills paths, - integrate AgentCore Memory, Gateway, MCP tools, Identity, Observability, Browser, or Code Interpreter, - review AgentCore security, least privilege, policy, tool exposure, credential handling, migration, or production readiness. ## Lean operating rules - First decide whether the user has an existing agent or needs a new project. Do not scaffold over an existing codebase. - For new projects, prefer the npm AgentCore CLI package `@aws/agentcore` because current AWS documentation recommends it. - Treat the Python starter toolkit as legacy/migration-oriented unless the user is explicitly working inside an existing Python-based toolkit workflow. - Separate code-based agents from config-based harnesses. Do not mix their guidance casually; current AWS docs describe the harness path as preview. - Prefer `AwsDocumentationMcpServer` when available via `uvx awslabs.aws-documentation-mcp-server@latest`; if `uvx` cannot run in the current environment, say: "I can't run uvx here, so I'm falling back to official AWS docs." Then