The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands using the acli CLI tool to interact with Jira Cloud services (SKILL.md).
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it retrieves and processes untrusted data from Jira summaries, descriptions, and comments.
Ingestion points: External data is read into the agent context via acli jira workitem view, acli jira workitem search, and acli jira workitem comment list in SKILL.md.
Boundary markers: The instructions lack markers or delimiters to help the agent distinguish between its own instructions and the data retrieved from Jira.
Capability inventory: The skill allows the agent to create/modify tickets, transition statuses, and execute shell commands, providing a significant action surface for injected instructions.
Sanitization: No sanitization or validation logic is specified for the data retrieved from external Jira fields.
[DATA_EXFILTRATION]: The acli jira workitem create command includes a --description-file parameter. This allows the agent to read the content of a local file and upload it as a Jira ticket description. This capability could be exploited to exfiltrate sensitive local files if the agent is manipulated into reading them and attaching the content to a public or accessible Jira ticket.

acli-jira