The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the datahub CLI tool to perform discovery and answer questions. It includes a specific security instruction to sanitize user input by rejecting shell metacharacters (such as backticks, dollar signs, and pipes) before passing it to the CLI.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted metadata from an external DataHub server, which presents an attack surface for indirect prompt injection.
Ingestion points: Metadata fetched via datahub search and datahub get operations (documented in SKILL.md).
Boundary markers: The skill does not implement explicit delimiters or instructions to the agent to ignore potentially malicious content within the retrieved metadata.
Capability inventory: The skill has access to the Bash tool to execute datahub commands.
Sanitization: While the skill provides logic to sanitize user-provided queries for shell injection, it does not verify or sanitize the content returned from the DataHub catalog (such as entity descriptions) for embedded instructions.

datahub-search