The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user input to generate GitHub Issue titles and bodies, creating a potential vector for malicious instructions to be embedded in data.
Ingestion points: Task summaries, background information, and acceptance criteria provided by the user in Step 1 are processed to generate CLI arguments.
Boundary markers: The instructions do not define specific delimiters or "ignore" instructions to separate user input from system prompts during processing.
Capability inventory: The skill has the ability to create and edit issues (gh issue create, gh issue edit) and perform arbitrary API requests (gh api) to the repository.
Sanitization: The skill relies on the model to correctly parse input into Conventional Commits format, but lacks explicit sanitization or validation of the resulting strings before they are executed in shell commands.
[COMMAND_EXECUTION]: The skill relies on the execution of shell commands through the GitHub CLI to interact with remote repositories.
Evidence: Frequent use of gh issue create, gh issue edit, and gh api to manage milestones and issue relationships. These commands target GitHub's official services, which are well-known and trusted platforms.

create-issue