The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from Jira issue summaries, descriptions, and comments, which could potentially contain malicious instructions aimed at influencing the agent's behavior.
Ingestion points: Untrusted data enters the agent context through issue view operations (jira issue view) and search results (mcp__atlassian__searchJiraIssuesUsingJql), as documented in SKILL.md and references/mcp.md.
Boundary markers: The skill includes instructions to "Show current vs proposed changes" and "Get approval before updating" in SKILL.md to mitigate accidental obedience to embedded instructions.
Capability inventory: The skill has the capability to execute shell commands via the jira CLI and perform Atlassian MCP tool calls for ticket creation, updates, and transitions.
Sanitization: There is no evidence of explicit sanitization or filtering of Jira ticket content before it is processed or displayed by the agent.
[COMMAND_EXECUTION]: The skill is designed to interact with the local environment by executing the jira CLI tool for issue management and sprint operations, as detailed in SKILL.md and references/commands.md.
[EXTERNAL_DOWNLOADS]: The documentation in references/mcp.md instructs the agent to execute a script located at ~/.claude/skills/jira/jira-link-issues to handle issue linking. This script is not included in the provided skill files, making its behavior unverifiable during this analysis.
[EXTERNAL_DOWNLOADS]: The skill recommends the installation of a third-party CLI tool (ankitpokhrel/jira-cli) via Homebrew or GitHub releases to enable core functionality.

jira