The Agent Skills Directory

[COMMAND_EXECUTION]: The skill dynamically generates a helper shell script at /tmp/jira-curl and modifies its permissions using chmod +x. This script acts as a wrapper for all subsequent curl operations to the Jira API.
[DATA_EXFILTRATION]: While the skill uses environment variables for the JIRA_API_TOKEN, the helper script is stored in /tmp. On multi-user or shared systems, the use of predictable paths in temporary directories for scripts handling authentication headers can lead to local credential exposure or session hijacking if not properly scoped.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its core functionality of ingesting external data from Jira.
Ingestion points: Fetches issue summaries, descriptions, and comments from external Jira projects via API endpoints.
Boundary markers: There are no explicit delimiters or 'ignore' instructions provided when the agent processes retrieved ticket content.
Capability inventory: The skill possesses file-writing capabilities, network access via curl, and the ability to execute generated scripts.
Sanitization: The data fetched from Jira (which could be controlled by an external attacker adding comments to a ticket) is passed to the agent without sanitization, potentially allowing malicious instructions in a ticket to influence the agent's behavior.

jira