The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands extensively to manage the software development lifecycle, including git operations for staging fixes, committing changes, pushing to remote repositories, and managing worktrees. It also utilizes the GitHub CLI (gh) to automate pull request tasks such as listing, marking ready, and merging pull requests.\n- [COMMAND_EXECUTION]: The verify-work.md reference describes a workflow that retrieves and executes custom shell commands defined in the project's .planning/config.json under the workflows.verify-work.extra_verification_commands key.\n- [DYNAMIC_EXECUTION]: The skill employs eval within its shell scripts and subagent instructions to execute command strings dynamically. This is notably used to run the user-provided extra verification commands sourced from the local project configuration.\n- [INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection because it ingests untrusted external data—such as project summary files and user-provided natural language feedback—and includes this data in the prompts used to spawn subagents for planning and checking fixes. This is a low-severity risk typical of LLM-based tools operating on user-influenced content, and the skill uses delimited blocks to provide context to subagents.

kata-verify-work