The Agent Skills Directory

[SAFE]: The skill and its associated scripts follow secure development practices. All operations are local to the agent environment, and the provided Python utilities (jql_query_builder.py and workflow_validator.py) use standard libraries without network or dynamic execution capabilities.
[INDIRECT_PROMPT_INJECTION]: The skill provides mechanisms to ingest natural language for JQL generation and JSON for workflow validation. This represents a potential attack surface for indirect prompt injection; however, the skill mitigates this by using structured mapping and logic-based validation scripts rather than unsafe string interpolation. 1. Ingestion points: jql_query_builder.py (natural language descriptions) and workflow_validator.py (workflow JSON definitions). 2. Boundary markers: The scripts act as transformation layers, effectively sanitizing input by mapping it to predefined JQL fragments or validating it against logical rules. 3. Capability inventory: Uses the mcp jira tool for searching, creating, and updating Jira issues and projects. 4. Sanitization: Input is parsed via regex and validated against a schema or library of safe patterns.

jira-expert