The Agent Skills Directory

[PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection.
Ingestion points: The jira-as ops discover-project command ingests untrusted project metadata and workflows from external JIRA instances.
Boundary markers: There are no instructions or delimiters defined to separate this external data from the system prompt.
Capability inventory: The skill is allowed to use Bash, Read, Glob, and Grep, which provides a broad execution environment for instructions processed by the agent.
Sanitization: No sanitization or validation of the ingested JIRA metadata is specified before it is utilized by the agent.

jira-operations