The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands to gather data from the local environment and remote sources. It uses git diff and git ls-files within loops to scan workspace directories and employs the GitHub CLI (gh pr diff and gh pr view) to fetch content from user-provided Pull Request URLs.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection by processing untrusted data from external specifications and using that data to perform actions in a corporate Jira environment.
Ingestion points: The skill reads full file content and diffs from .ai/spec/ files located in external GitHub Pull Requests via the gh tool and local git commands.
Boundary markers: There are no explicit instructions or delimiters provided to the superpowers:brainstorming tool to treat the specification content as untrusted data or to ignore instructions embedded within those files.
Capability inventory: The skill has extensive write permissions in Jira, including createJiraIssue, editJiraIssue, transitionJiraIssue, and addCommentToJiraIssue across Epics and Stories.
Sanitization: While the skill strips JQL reserved characters (-, (, ), etc.) when constructing search queries to prevent JQL injection, it does not perform sanitization on the specification text before it is used to generate Jira summaries, descriptions, and acceptance criteria.

make-jira-from-spec