The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill documents patterns for processing untrusted data from various AWS event sources.
Ingestion points: Lambda event objects originating from API Gateway, S3, SNS, and SQS (documented in references/raw-python-lambda.md and references/chalice-lambda.md).
Boundary markers: The skill recommends using Pydantic for data validation and explicitly warns to "always validate and sanitize event data" in the constraints section of SKILL.md.
Capability inventory: The provided code examples perform standard operations such as DynamoDB writes, S3 uploads, and logging. No unsafe usage of subprocesses or dynamic code execution on user input is present.
Sanitization: The instructions emphasize input validation and structured logging to mitigate injection risks.
[EXTERNAL_DOWNLOADS]: The skill includes instructions for installing standard development dependencies and deployment tools.
Evidence: Documentation in references/serverless-deployment.md and references/testing-lambda.md provides commands for pip install and npm install targeting well-known community packages like chalice, serverless, moto, and pytest.
Context: These are standard development workflows for the intended use-case of building and testing serverless applications.
[COMMAND_EXECUTION]: The skill demonstrates the use of shell-based build scripts and CI/CD workflows.
Evidence: references/raw-python-lambda.md includes a build.sh script for packaging Lambda functions, and references/serverless-deployment.md provides GitHub Actions YAML configurations executing SAM and Serverless CLI commands.
Context: The execution is restricted to standard build and deployment automation using the author's provided templates.

aws-lambda-python-integration