The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to Indirect Prompt Injection because it is designed to ingest and process content from external Jira tickets.
Ingestion points: Untrusted data enters the agent context through issue titles, descriptions, and comments retrieved via CLI and MCP tool operations (referenced in SKILL.md).
Boundary markers: SKILL.md provides instructions to treat Jira content as untrusted input and recommends using specific patterns (e.g., quoting, /tmp file input) to separate data from commands.
Capability inventory: The skill provides the agent with capabilities to execute shell commands via the jira CLI and perform broad Jira modifications using MCP tools.
Sanitization: The instructions explicitly warn the agent against direct interpolation of ticket content into shell strings, though enforcement relies on model compliance with the provided guidelines.
[EXTERNAL_DOWNLOADS]: The skill recommends the installation of the jira-cli tool from the ankitpokhrel/jira-cli repository on GitHub. This is a well-known community tool for Jira interaction.
[PROMPT_INJECTION]: The SAFETY.md file contains a security evaluation report and the skill name uses 'hardened' as a prefix. These represent self-referential safety claims that must be treated as informational data rather than authoritative proof of safety during analysis.

jira-hardened