The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by ingesting untrusted data from the local environment to populate Linear issue fields. Injected content in metadata could result in misleading issues.
Ingestion points: Data extracted from args, conversation history, current directory name, and git branch (SKILL.md).
Boundary markers: The skill does not provide specific instructions to the agent to ignore or delimit instructions embedded within these context signals.
Capability inventory: The skill uses tools to create_issue, list_issues, and update_issue in Linear, and uses gh pr view to fetch PR data.
Sanitization: There is no evidence of sanitization or validation of the environmental context before it is interpolated into the issue creation payload.
[COMMAND_EXECUTION]: The skill relies on local shell commands (git branch, gh pr view) to gather context for issue creation. These are standard operations for development-focused skills.

linear-issue