The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses potentially dangerous commands such as rm -rf to delete local directories and cf undeploy --delete-services to remove remote infrastructure. While these are central to the skill's purpose, they represent a significant impact if misused. The skill attempts to mitigate this with 'HARD RULES' requiring explicit user confirmation for every destructive step.
[INDIRECT_PROMPT_INJECTION]: The skill reads the ID field from mta.yaml and the output folder name, then interpolates these values directly into shell commands (e.g., cf undeploy <mta-id> and rm -rf <output-folder>).
Ingestion points: The skill parses mta.yaml using grep to extract identifiers (SKILL.md).
Boundary markers: No explicit boundary markers or sanitization logic is provided to handle special characters or command separators in the extracted strings.
Capability inventory: The skill has access to the Bash tool and performs file deletion and remote infrastructure teardown.
Sanitization: There is no evidence of input validation or escaping for the strings extracted from local files before they are used in shell commands. A maliciously crafted mta.yaml could attempt command injection.

appfront-cleanup