The Agent Skills Directory

[COMMAND_EXECUTION]: In references/commands.md, the skill recommends a pattern for creating issues with multi-line descriptions using a shell heredoc (cat > /tmp/jira_body.md <<'EOF'). This pattern is vulnerable to command injection if the input content, which may come from untrusted sources like Jira tickets or user-provided descriptions, contains the delimiter string EOF. An attacker could use this to terminate the heredoc prematurely and execute arbitrary shell commands in the agent's environment.
[COMMAND_EXECUTION]: The documentation in references/mcp.md instructs the agent to execute a script located at ~/.claude/skills/jira/jira-link-issues. This script is described as 'bundled' but is not present in the provided skill files, creating a dependency on an unverified and missing executable.
[EXTERNAL_DOWNLOADS]: The skill suggests downloading and installing the jira-cli tool from a public GitHub repository (ankitpokhrel/jira-cli). While this is a widely used community tool, users should verify its integrity before installation as it is not a first-party binary.
[SAFE]: The skill follows security best practices by using environment variables (JIRA_API_TOKEN) for credential management and explicitly instructing the agent to request user approval before any modification operations.

jira