The Agent Skills Directory

[COMMAND_EXECUTION]: The skill constructs and executes a PowerShell command string that incorporates several user-provided placeholders, such as the path to the SSMA executable ({SSMA_CONSOLE_PATH}), the script XML, and the operation name. This pattern can be vulnerable to command injection if the inputs contain malicious shell characters.
[CREDENTIALS_UNSAFE]: To perform the migration, the skill collects Oracle and SQL Server passwords and writes them in cleartext into XML configuration files (ssma-variables.xml and ssma-servers.xml) on the local file system. While this is a requirement for the SSMA console tool, storing secrets in plain text on disk poses a significant security risk if the environment is compromised.
[INDIRECT_PROMPT_INJECTION]: The skill is instructed to read generated migration reports and logs to summarize findings. If the source database metadata (e.g., table or schema names) contains malicious instructions, they could be reflected in these reports and potentially influence the agent's summary or subsequent actions.

ssma-console