The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on git, gh (GitHub CLI), and gt (Graphite CLI) to automate branch management, PR synchronization, and pushing changes. These commands are executed locally to perform the defined stack-merge workflow.
[PRIVILEGE_ESCALATION]: The skill uses administrative bypass capabilities for repository branch protection to enable fast-forward pushes to the main branch. It performs explicit read-only checks (in scripts/validate.sh) of the user's repository permissions and the branch's protection status to ensure the operation is authorized and safe to perform.
[INDIRECT_PROMPT_INJECTION]: The skill processes external inputs such as branch names, commit messages, and pull request metadata.
Ingestion points: Data is ingested via gh pr view, gh pr list, and gt ls commands.
Boundary markers: While the skill defines distinct operational phases, it does not use explicit boundary delimiters for the interpolated branch data.
Capability inventory: The skill can perform force-pushes, update the main branch, and close pull requests.
Sanitization: Shell scripts use secure argument handling and quoting (e.g., using shell arrays for branch lists) to mitigate command injection risks from external data like branch names.
[DYNAMIC_EXECUTION]: The skill utilizes python3 -c for parsing JSON output from the GitHub CLI. This execution is limited to a static, localized command for data formatting and does not incorporate untrusted code or remote payloads.

stack-merge