The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes several internal Python scripts (e.g., scoring.py, model_recommendation.py, score_units.py) using uv run to perform deterministic analysis. These scripts are part of the skill's distribution and are used for intended functional logic.
[EXTERNAL_DOWNLOADS]: The skill makes requests to well-known domains like docs.temporal.io and aws.amazon.com for documentation freshness checks. It also utilizes AWS Model Cards and pricing data. These are legitimate resources for an architecture advisor.
[DYNAMIC_EXECUTION]: The skill implements a custom DSL (defined in INTERPRETER.md) to manage execution flow across different markdown files. While this is a form of dynamic instruction loading, it is restricted to the skill's own internal file structure.
[DATA_EXPOSURE]: The POC generation and model verification phases involve standard AWS CLI and SDK calls (e.g., sts get-caller-identity, Bedrock model invocation). These operations are gated by user confirmation and are necessary for the skill's stated purpose of verifying infrastructure and model access.

agent-advisor