The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data that could influence agent behavior.
Ingestion points: The ISSUE_PROBLEM variable (extracted from an external issue file in Step 1.5) and the DESCRIPTION variable (sourced from user input in Step 2) are interpolated into prompts for sub-agents.
Boundary markers: The skill uses XML-style tags (e.g., <planning_context>, <agent-instructions>) to delimit data in prompts, which provides some structure but does not include explicit warnings for sub-agents to ignore embedded instructions within the ingested content.
Capability inventory: The skill has significant capabilities, including filesystem modification, Git operations (commit/push), and GitHub interaction via the gh CLI.
Sanitization: There is no evidence of sanitization or escaping of the external issue content or user input before it is passed to the planner and executor sub-agents.
[COMMAND_EXECUTION]: The skill uses local shell commands for orchestration logic, filesystem management, and source control.
It performs atomic commits and manages project state files using standard bash utilities (sed, grep, xargs, git).
It utilizes the GitHub CLI (gh) to automate Pull Request creation and issue management.
[EXTERNAL_DOWNLOADS]: The skill interacts with remote services as part of its development workflow.
It performs git push and uses gh pr create to send data to GitHub. These operations target a well-known service (GitHub) and are consistent with the skill's purpose of managing development tasks.

kata-execute-quick-task