The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses spawnSync to execute various CLI tools, including git for branch and remote management, gh (GitHub CLI) for PR creation, and glab (GitLab CLI) for merge request management. It also executes utility commands like which and readlink to locate local installation paths for related tools.
[CREDENTIALS_UNSAFE]: To authenticate with the Jira API, the scripts read sensitive configuration files from the local filesystem, such as ~/.config/.jira/.config.yml (standard go-jira configuration) and a .jira-token file. This access is required for the skill to perform operations on behalf of the user.
[DATA_EXFILTRATION]: The skill transmits repository metadata, pull request URLs, and authentication tokens to the configured Jira server and code hosting platforms. By default, it communicates with an official Jira instance at redhat.atlassian.net and standard GitHub/GitLab domains.
[EXTERNAL_DOWNLOADS]: The scripts perform network operations using the fetch API and CLI tools to interact with Jira, GitHub, and GitLab servers.
[INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it processes untrusted data such as PR titles and descriptions.
Ingestion points: Command-line arguments (--title, --body) in scripts/create-pr-mr.js and scripts/link-pr-mr.js accept arbitrary text from the agent session.
Boundary markers: No explicit delimiters are used to wrap the PR body before it is passed to CLI tools or the Jira API.
Capability inventory: The skill possesses file-read access (credentials), command execution capabilities, and network access.
Sanitization: The use of spawnSync with argument arrays provides a baseline level of protection against shell command injection, though the content itself remains unsanitized before being sent to external APIs.

rhdh-jira-link