The Agent Skills Directory

[SAFE]: The skill serves as legitimate documentation for AWS Bedrock. No malicious behaviors, obfuscation, or unauthorized access patterns were identified. External references and dependencies are limited to well-known technology services and official repositories.
[COMMAND_EXECUTION]: The skill provides numerous examples of AWS CLI commands (e.g., aws bedrock, aws bedrock-runtime, aws bedrock-agent-runtime) for invoking and managing foundation models. These commands are standard for the described tasks.
[EXTERNAL_DOWNLOADS]: Scripts utilize the official boto3 and botocore libraries to communicate with Amazon Bedrock API endpoints, which is required for the functionality described.
[INDIRECT_PROMPT_INJECTION]: The skill provides code templates for applications that ingest user input for model invocation. 1. Ingestion points: Python functions such as invoke_claude and the Conversation class accept user messages as parameters. 2. Boundary markers: The examples utilize standard JSON message structures with defined roles and explicitly highlight the implementation of Bedrock Guardrails for content filtering. 3. Capability inventory: The provided code utilizes network-based API calls to AWS services via boto3. 4. Sanitization: The skill documents the use of Bedrock Guardrails to sanitize and filter both inputs and outputs.

bedrock