The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill presents an exposure to indirect prompt injection due to processing untrusted data alongside active system capabilities.
Ingestion points: In SKILL.md, references/commands.md, and references/mcp.md, the skill reads issue data, comments, and descriptions via tools like mcp__atlassian__getJiraIssue and commands like jira issue view.
Boundary markers: Absent. No protective isolation delimiters or instructions are used to shield the agent from instructions embedded in issue text.
Capability inventory: The skill utilizes local shell commands (jira CLI) and network operations (Atlassian MCP) to create, modify, and delete tickets.
Sanitization: Absent. No filtering or escaping mechanisms are specified for incoming Jira ticket data.
[COMMAND_EXECUTION]: The skill instructs the agent to run shell commands via the jira CLI and a bundled script named jira-link-issues to manage project workflows and tickets locally, as shown in SKILL.md and references/commands.md.

jira