The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches and executes the aws-diagram-mcp-server utility from the awslabs organization using the uvx package runner to handle infrastructure scanning and diagram generation.
[COMMAND_EXECUTION]: Executes shell commands to run the MCP server using uvx and provides instructions for installing the graphviz system dependency via package managers like apt or brew.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the AWS environment, such as resource names, tags, and metadata, which could potentially contain malicious instructions targeting the agent during diagram generation.
Ingestion points: Infrastructure metadata retrieved from live AWS APIs, including resource tags and configuration values (SKILL.md).
Boundary markers: No specific delimiters or warnings are defined to separate ingested resource data from instructions.
Capability inventory: Executes uvx for tool startup, uses graphviz for file rendering, and performs network requests to AWS service endpoints.
Sanitization: The skill does not document any sanitization or escaping procedures for metadata before it is processed for diagram output.

aws-architecture-diagram