The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill manages Jira API credentials, reading them from environment variables (JIRA_API_TOKEN) or local configuration files (~/.config/.jira/.config.yml, .jira-token). This behavior is standard for the skill's documented purpose of interacting with the Jira API.
[COMMAND_EXECUTION]: The scripts execute several local command-line utilities including git, gh (GitHub CLI), and glab (GitLab CLI) to manage repositories and PR/MR creation. These operations are performed using spawnSync with argument arrays, which is a secure method to prevent shell injection.
[EXTERNAL_DOWNLOADS]: The skill makes network requests to Jira API endpoints (atlassian.net) and GitLab/GitHub instances to fetch issue details and manage remote links. These interactions are necessary for the skill's primary functionality.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it processes untrusted data such as PR titles and Jira issue summaries.
Ingestion points: Data enters the agent context through the --title and --body arguments in create-pr-mr.js and via Jira issue fields fetched in link-pr-mr.js (scripts/create-pr-mr.js, scripts/link-pr-mr.js).
Boundary markers: The documentation recommends using /prose-editing for input processing, but no explicit programmatic delimiters are enforced within the scripts themselves.
Capability inventory: The skill has the capability to execute shell commands (git, gh, glab) and perform network operations via fetch (Jira API).
Sanitization: Shell injection is mitigated by using argument arrays in spawnSync, though the semantic content of the inputs is not validated for malicious instructions.

rhdh-jira-link