The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is centered around executing commands via the ae-cli binary to manage external automation resources.
Evidence: The skill defines a complete command contract in references/cli.md for operations such as ae-cli automation create, update, and delete.
Mitigations: Instructions in SKILL.md and references/cli.md explicitly require user confirmation for high-risk operations (deleting/stopping) and forbid the use of the --yes flag unless the user has already authorized the action.
[INDIRECT_PROMPT_INJECTION]: The skill provides a surface for indirect injection by interpolating untrusted user data into shell command arguments.
Ingestion points: User-provided values for --name NAME and --message TEXT in the ae-cli automation create command (documented in references/cli.md).
Boundary markers: The instructions do not define specific delimiters or escaping mechanisms to prevent these strings from being used for command injection or subverting the CLI's intended behavior.
Capability inventory: The skill has the ability to perform persistent writes, trigger task executions, and delete resources.
Sanitization: There is no mention of sanitization or validation of the input strings before they are passed to the CLI tool.

ae-automation