The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the jira command-line tool (specifically ankitpokhrel/jira-cli) to interact with Jira. It includes safety checks to verify the tool's presence and version before execution.
[CREDENTIALS_UNSAFE]: The skill mentions handling JIRA_API_TOKEN but provides robust safety guidelines. It explicitly forbids the agent from printing secrets, persisting them, or placing them in command arguments, recommending secure alternatives like environment variables or keychains.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data when reading Jira issues and comments.
Ingestion points: Commands such as jira issue view and jira issue list (in SKILL.md and references/commands.md) ingest content from external Jira instances.
Boundary markers: The skill encourages using structured output flags like --plain, --raw, or --csv to facilitate safer parsing.
Capability inventory: The agent has the ability to write to the Jira API (mutations) and read external content, but these are gated by authorization requirements.
Sanitization: Instructions mandate quoting values and using stdin or templates for long content to prevent shell injection when handling external data.
[SAFE]: The skill implements a strict authorization workflow for all destructive or additive operations, requiring the agent to show the proposed mutation and receive explicit user approval before execution.

use-jira-cli