The Agent Skills Directory

[Command Execution]: The skill executes shell scripts located in the local project directory (.claude/scripts/worktree-create.sh and .claude/hooks/tdd-state.sh) and uses standard command-line utilities such as git and the GitHub CLI (gh). These operations are used to manage repository state and facilitate the creation of stacked Pull Requests, which align with the skill's stated goal.
[Data Ingestion and Processing]: The skill retrieves information from external sources (GitHub issues) to guide the behavior of subagents.
Ingestion points: GitHub issue descriptions are fetched in Step 2 and Step 5 to extract requirements and acceptance criteria.
Boundary markers: The skill does not explicitly define delimiters for external data, relying on the subagents' prompts to interpret the issue content.
Capability inventory: The skill possesses the ability to create worktrees, perform git rebases, push to remote repositories, and create Pull Requests.
Sanitization: The skill includes a step to sanitize issue numbers (Step 1) by removing punctuation and whitespace, reducing the risk of command injection through issue identifiers.
[Environment Awareness]: The skill checks for the environment variable CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS to determine if it should proceed with full automation or fall back to a setup-only mode. This is a standard practice for managing experimental features.

sprint