The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/setup script automates the installation of system dependencies.
It installs jq and ankitpokhrel/jira-cli/jira-cli using the Homebrew package manager if they are not already present in the environment.
[INDIRECT_PROMPT_INJECTION]: The skill processes content retrieved from external Jira issues and Confluence pages, creating a surface for indirect prompt injection.
Ingestion points: Data enters the agent's context through jira issue view, jira issue list, confluence get, confluence search, and confluence comments subcommands.
Boundary markers: The instructions do not define explicit delimiters or instructions to ignore embedded commands within the retrieved Atlassian content.
Capability inventory: The agent has the ability to create and modify Jira issues, create and update Confluence pages, and execute shell commands through the allowed tools.
Sanitization: The skill uses jq with --arg and --argjson flags to safely construct JSON payloads and applies URI encoding to search queries, which mitigates basic injection into API requests.
[CREDENTIALS_UNSAFE]: The skill manages sensitive Atlassian Cloud API tokens and account emails.
Storage: Credentials are saved to ~/.atlassian-env.
Mitigation: The setup script explicitly sets the file permissions to 600 (read/write by owner only) and automatically appends the file name to the user's global .gitignore (via core.excludesfile) to prevent accidental version control commits.

atlassian